Introduction to UMTS Optimization

8/13/2019 Introduction to UMTS Optimization

http://slidepdf.com/reader/full/introduction-to-umts-optimization 1/429

Introduction toUMTS OptimizationCourse Code: SC2804 Duration: 2 days Technical Level: 3

Radio Principles and Planning courses include:

Radio Principles

Principles of Radio Site Engineering

Digital Radio and Microwave Link Planning

Cell Planning for GSM Networks

2G/3G Indoor Coverage Planning

3G Indoor Coverage Planning

Introduction to GSM Optimization

Drive-Test Data Capture and Analysis

Cell Planning for UMTS Networks

Introduction to UMTS Optimization

Specially prepared for Safaricom Limited





© Wray Castle Limited

INTRODUCTION TO UMTS

OPTIMIZATION

First published 2004Last updated October 2008WRAY CASTLE LIMITED

BRIDGE MILLSSTRAMONGATE KENDAL

LA9 4UB UK

Yours to have and to hold but not to copy

The manual you are reading is protected by copyright law. This means that Wray Castle Limited could take you and

your employer to court and claim heavy legal damages.

Apart from fair dealing for the purposes of research or private study, as permitted under the Copyright, Designs andPatents Act 1988, this manual may only be reproduced or transmitted in any form or by any means with the prior

permission in writing of Wray Castle Limited.




© Wray Castle Limitedii





Section 1 Introduction and Overview

Section 2 Optimization Software Tools

Section 3 Optimizing Coverage and Capacity

Section 4 RAN Configurations and Dimensioning

Section 5 Idle Mode and System Access

Section 6 Connected Mode and Radio Link Control

Section 7 UMTS Features and Techniques

INTRODUCTION TO UMTS OPTIMIZATION

CONTENTS

iii




© Wray Castle Limitediv





SECTION 1

INTRODUCTION AND OVERVIEW

v




© Wray Castle Limitedvi





1 Optimization or Planning? 1.11.1 What is Optimization? 1.11.2 Typical Planning/Optimization Distinction 1.31.3 Differences for UMTS 1.5

2 The Optimization Process 1.7

2.1 Identifying Optimization Opportunities 1.72.2 Key Statistics and Analysis 1.72.3 Drive Tests and Signalling Analysis 1.72.4 Change Implementation 1.92.5 Monitoring 1.112.6 Database Update 1.11

3 Exercise 1 – Discussion about Optimization Optionsand Priorities 1.13

4 Drivers for Optimization 1.15

4.1 Overall Quality of Service (QoS) 1.154.2 Set-up Failure 1.174.3 Dropped Calls 1.19

5 The Coverage–Capacity–Quality Relationship 1.215.1 Interference Sources 1.215.2 The Coverage Loop 1.23

6 Summary of Optimization Strategies 1.25

CONTENTS

vii




© Wray Castle Limitedviii





At the end of this section you will be able to:

• explain the close relationship between planning and optimization in a

Wideband CDMA (WCDMA) radio network

• describe the overall optimization process as distinct from purely planning

functions

• list typical key metrics relating to optimization

• outline, in general terms, how the air interface may be optimized through the

use of cell parameters, activation of features and other techniques

OBJECTIVES

ix




SC2804/S1/v1.11.1 © Wray Castle Limited

1.1 What is Optimization?

The term optimization is used in connection with almost any engineering design task.It is usually taken to mean fine tuning for optimum performance. This generalunderstanding of the term can be applied comfortably in the context of a UMTSnetwork, but its precise interpretation can vary a great deal in practice.

Ideally, the optimization of a Universal Mobile Telecommunications System (UMTS)network would take place in the assumption that the network is not under performingbecause of some fault condition or configuration error. In practice, however, theoutput of the optimization process will often be the identification of a fault or incorrectly-set parameter value.

The optimization process may also stray from its purest interpretation into the area of future planning. The nature of UMTS network design is such that it benefits fromgiving consideration to future direction even when planning for current needs. Theoptimization team is in a good position to estimate the likely future behaviour of thenetwork and may provide a valuable input into future planning needs.

1 OPTIMIZATION OR PLANNING?




1.2© Wray Castle Limited

fine tuning for optimum performance

Optimization – Theoretical

fine tuning for optimum performance

fault/configuration error detection

identification of network development requirements

setting planning goals

Optimization – Practical

Figure 1

Optimization Definition

SC2804/S1/v1.1





1.2 Typical Planning/Optimization Distinction

Most people distinguish between the planning and optimization processes. This istrue whatever the technology because it would be impossible to perform any kind of optimization on a network that had not yet been planned. Therefore, planning can beconsidered as a process that is carried out and completed before optimizationcommences. Furthermore, the optimization process will need a goal, for example acertain minimum level of dropped calls. Therefore it also makes sense to consider that until a network’s performance can be observed and judged, it cannot beoptimized. This idea emphasizes a division in time between planning andoptimization.

Much of this is true of the Global System for Mobile communications (GSM). TheGSM planning process is generally one of ensuring sufficient radio coverage basedon assumptions made in formulating link budgets. The process of coverage planningcan be independent of capacity planning. This means that the initial planning processcan be performed without optimization involvement.





Set targets forradio coverageand capacity

Perform linkbudget calculations

and planning forradio coverage

Dimension forcapacity

requirements

Build the network

Gatherperformance

statistics

Plan for

continuednetwork

development

Optimize radionetwork design

and configuration

Figure 2

Planning and Optimization Relationship in GSM

SC2804/S1/v1.1





1.3 Differences for UMTS

For UMTS, coverage and capacity planning must be linked. This is because themutual interference between calls has a direct impact on radio performance, henceon coverage.

This means that even at the earliest stage a proposed radio network design shouldbe tested, evaluated and optimized in traffic-loaded conditions. The only way to dothis at the design stage is by simulation. A realistic and detailed simulation will bebeneficial. Similarly, the earlier the optimization process can be carried out the better.

This can be thought of as ‘optimization in advance’. However, no simulation isperfect and traffic characteristics can only be guessed. This means that constantmodification is required as the real network is rolled out and real trafficcharacteristics become apparent. In UMTS, planning and optimization are ongoingprocesses that will always remain closely linked.





Set targets forradio coverageand capacity

Perform linkbudgets and

traffic analysis todetermine cellcharacteristics

and configuration

Plan radionetwork including

expected

expansion afterrollout

Build the network

Gatherperformance

statistics

Plan for continuednetwork

development

Optimize radionetwork design

and configuration

Optimize throughsimulation

Optimize throughsimulation

Figure 3

Planning and Optimization Relationship in UMTS

SC2804/S1/v1.1





2.1 Identifying Optimization Opportunities

The first step is to distinguish between optimization problems and faults andconfiguration problems. Information is therefore required from a number of sources,for example:

• key performance statistics

• problem reports from customers

• radio planning information

• recent configuration changes

• completed and ongoing work

• existing data on problem areas

Analyzing this data and correlating the information will enable true optimizationopportunities to be identified.

2.2 Key Statistics and Analysis

The next step is statistical analysis of all the sites with an optimization problem.

Radio planning will give information about anticipated problems such as interferenceand coverage. Historical data on previous problems may indicate a new issue hasarisen, perhaps due to expansion or an increase in load factor on one or more cells.

There may now be enough information to suggest a solution. If not, further information may be obtained by drive testing.

2.3 Drive Tests and Signalling Analysis

Performing a drive test in the area where the problem exists may result in further data. Failing that, detailed analysis of the signalling information passed between theNode Bs and Radio Network Controllers (RNC) may uncover the problem.

To make the drive test, call trace and signalling measurements valid they should beperformed under the same conditions as those prevailing when the original problemoccurred. For example, at the same time of day, in the same traffic conditions, on thesame route and in the same place.

2 THE OPTIMIZATION PROCESS





Inputs:QoS targets, problem reports,planning information, ongoing work

Inputs:radio planning, historical data

No

Yes

Yes

No

Identifying anoptimization opportunity

Statistical analysisof all sites of interest

Identify anappropriate change

Implement change

Monitor results

Update database

Sufficient information

Success Reverse change

Perform drive test

Figure 4

The Optimization Process

SC2804/S1/v1.1









No

Yes

Yes

No




Implement change

Monitor results

Update database



Perform drive test

Figure 4

The Optimization Process (repeated)

SC2804/S1/v1.1





2.5 Monitoring

Having made the change it is important to perform post-implementation monitoring toensure it has the desired effect. This can be done by monitoring the statistics or,better still, by using the same method as was used to identify the problem initially.Statistical analysis should also be carried out to assess the impact, if any, on the restof the network. In UMTS this monitoring must include observation of surroundingcells.

2.6 Database Update

If the changes have been successful (or not), the databases in the networkmanagement systems need to be updated. This way the history of the problem, andhopefully its solution, can be logged and used by others.







No

Yes

Yes

No




Implement change

Monitor results

Update database



Perform drive test

Figure 4

The Optimization Process (repeated)

SC2804/S1/v1.1





Working in groups of two or three, complete the following exercise and summarize

your group’s answers in the work space on the opposite page. Allow about 10minutes, after which all groups will compare answers.

1 List techniques, features or solutions that reduce interference either directly or indirectly (e.g. antenna downtilt).

2 List techniques, features or solutions that increase capacity either directly or indirectly (e.g. secondary scrambling codes).

3 List techniques, features or solutions that improve radio coverage or producebetter utilization of existing coverage (e.g. cell repeater).

4 List techniques, features or solutions that combat slow fading and fast fadingand their effects, either directly or indirectly (e.g. transmit diversity).

5 List techniques, features or solutions that improve link quality either directly or indirectly (e.g. multi-user detection).

3 EXERCISE 1 – DISCUSSION ABOUT OPTIMIZATION OPTIONS

AND PRIORITIES





Exercise 1

Work Space and Summary of Results

SC2804/S1/v1.1





4.1 Overall Quality of Service (QoS)

For a network to be successful in the highly competitive mobile phone market, it mustbe customer driven. This should be reflected in the setting of appropriate Quality of Service (QoS) targets against which network performance can be measured on aregular basis. The QoS targets must be reviewed regularly as part of a policy of constant improvement.

The UMTS standards associate a specific technical meaning to the term QoS indescribing the expected performance characteristics of a channel. These are valid inthis context, but the term is also being used in a wider sense. Here it includes acustomer’s personal feeling about the success and usability of a service. Thus itincludes what may be termed ‘human factors’.

Measurement of the QoS may be carried out either by the network operator or by anindependent agency or a combination of the two.

In terms of air interface performance for real-time services such as voice, customersare usually concerned primarily with call success rate and secondarily with callquality. For non-real-time services such as messaging or data exchange, thisprioritization may be reversed. Call success rate could be defined in a number of ways, but a simple definition classifies calls as successful when they set up without a

problem, do not suffer handover failure and clear normally, i.e. the call is not clearedabnormally or dropped. Given the slight differences in processes, it is wise tomeasure call success rate independently for mobile-terminated calls and for mobile-originated calls.

Call quality may be measured in a number of ways depending on the type of call.Voice or video may be judged subjectively, but for optimization purposes an objectivetarget in terms of bit error rate or frame erasure rate is preferable. Data andmessaging services can also be considered in terms of bit error rate and frame error rate, but a retransmission factor should also be considered. Data services will alsohave delay requirements in terms of latency and delay variation.

Finally, the quality of the radio channel may be a good indicator of overall quality andthis may be monitored in terms of radio signal strength and signal-to-noise ratio.

4 DRIVERS FOR OPTIMIZATION







4.2 Set-up Failure

Call set-up failure is attributable to a variety of causes. There may be a hardwareand/or software failure in the network or in the mobile equipment; alternatively, theUMTS Subscriber Identity Module (USIM) may be invalid or faulty.

In relation to the air interface, congestion may be the cause, possibly within theRandom Access Channel (RACH) or Paging Control Channel (PCCH). Generally thistype of congestion only affects mobile-terminated calls, but PCCH congestion mayalso affect some types of ongoing data calls.

The congestion of traffic-carrying channels will be a significant concern for optimizers. When the cell’s noise rise limit is reached, Radio Resource Control(RRC) will not allow new calls to be established. This situation in UMTS iscomplicated by the simultaneous provision of different service types with differentQoS requirements. For example, a real-time voice call or higher-bit-rate video callmay be blocked because of the noise rise limit. Yet, at the same time, a low-bit-ratenon-real-time call may be allowed to go ahead. Additionally, the noise rise in a cell tobe partly a factor of traffic load in neighbour cells, so it is possible for congestion inone cell is caused by overloading in a neighbour cell. Care must be taken to ensurethat the cause is the focus of optimization, not the symptom.

Calls may also fail at setup because of poor radio coverage, fading, or interferencecausing failure in access channels. Coverage can never be perfect. Interference isalways present and can become too strong. Fading effects are also inevitable in acluttered, multipath environment.

The most obvious sources of interference are other users and other intra-frequencycells. However, interference contributions will also be present from inter-frequencycells, some of which could belong to other operators. This may be an importantconsideration in some optimization scenarios.

The multimedia nature of Third-Generation (3G) services means that not all networks

will support all services in all locations. Therefore it is possible that calls may failsimply because the network does not support the requested service or channelconfiguration.

Incorrect cell parameter settings could also cause set-up failure, for example bycausing mobiles to select an inappropriate server in idle mode or use inappropriatetransmit power for access. UMTS presents particular challenges for the optimizer inthis respect because there are so many parameters and because of theinterdependency between cells.





Set-up Failure

congestion

poor radio coverage

interference

fading

service not supported

incorrect or suboptimal cell parameter settings

hardware/software problem in the network,mobile equipment or USIM

Hardware limitsSoft capacityService type and QoS variation Air interface channel types

Intra-frequencyInter-frequencyInter-operator Pilot pollutionExternal Noise

Many parametersInterdependency

Figure 6

Set-up Failure

SC2804/S1/v1.1





4.3 Dropped Calls

Many of the reasons why calls drop are closely related to those that cause set-upfailure. For example, calls may drop because of a hardware or a software problem inthe network or mobile equipment, or because of problems in the radio channel.

The potential causes of problems with the radio channel in terms of signal strengthor interference are the same as those for set-up failure. One additional problemwhen considering dedicated channels could be the inappropriate setting of parameters that relate to closed loop power control.

Calls requiring dedicated channels will also need handover functions. These may bea mixture of soft and hard handovers. In most UMTS networks there is also arequirement for inter Radio Access Technology (RAT) handovers. There are manyparameters that relate to measurements and subsequent handover decisions.Incorrect or inappropriate setting of these parameters could result in handover failure. Problems with coverage or interference could also result in handover failure.In extreme cases call drops may be forced on a priority basis at times of congestion.

If pre-emptive channel allocation is adopted for emergency (112) calls, then a routinenon-emergency call may be dropped to provide emergency capacity.

Key metrics relating to dropped calls include poor signal level, high interference leveland handover success/failure rate.





Dropped Calls

interference

fading

poor radio coverage

handover/reselection failure

fast power control

incorrect or suboptimal cell parameter settings

pre-emption for emergency call channelallocation

Different bit ratesDifferent QoS

Intra-frequencyInter-frequencyInter-operator Pilot pollutionExternal noise

CapacityQuality

Soft (intra-frequency)

Hard (inter-frequency)Hard (inter-RAT)

MeasurementsPower controlHandover

Figure 7

Dropped Calls

SC2804/S1/v1.1





5.1 Interference Sources

The capacity available in a UMTS system is ultimately limited by the amount of interference present. Downlink capacity may be thought of as limited by the totalamount of transmit power available from the Node B. Nonetheless, downlink transmitpower is only a factor because the inability to raise power beyond a limited pointrestricts the ability to overcome interference.

The amount of interference tolerated by a given system is variable. It can beconsidered a factor of three key considerations:

• services offered• features supported

• local environment

Different services have different QoS requirements and can therefore toleratedifferent amounts of interference. Optional features such as Multi-User Detection(MUD) can be used to increase tolerance to interference. The local environmentdetermines a channel’s exposure to potential interference sources.

An uplink channel is separated from other channels by uplink scrambling codes. An

individual channel will experience interference predominantly from other in-cell andneighbour-cell intra-frequency channels. However, there will also be some adjacentchannel interference, which may be most problematic if the interference sourcebelongs to another operator.

A downlink channel is separated from other channels on the same cell by theOrthogonal Variable Spreading Factor (OVSF) codes. These are highly orthogonal,but where different-length codes are used simultaneously in a multipath environmentthere will be a significant interference contribution. Downlink channels in neighbour cells are separated by scrambling codes, but this will also present an interferencesource. Additionally, as for uplink channels, adjacent radio channels will contributesome interference.

5 THE COVERAGE–CAPACITY–QUALITY RELATIONSHIP





Inter-frequencyneighbour

ServingNode B

UE

Intra-frequencyneighbour

DL

UL

Other UEs inneighbour cells

Other UEs inneighbour cells

Other UEs inthe serving cell

UL Int.

UL Int.UL Int.

DL Int.

DL Int.

DL Int.

Figure 8

Interference Sources

SC2804/S1/v1.1







Coverage

CapacityLink

Budget

Figure 9

The Coverage Loop

SC2804/S1/v1.1





Most optimization solutions involve the use of network features, adjustment of one or

more cell parameters, adjustment of antenna orientation, tilt, height or type, andredimensioning of traffic or control channels. More serious issues may require theaddition of macro or micro sites, provision of in-building coverage, or cell splitting.

In all cases, optimization activity must be carefully prioritized, keeping QoS and thecustomer in mind. There is little point in trying to optimize a cell working at 90% of potential capacity if one of its neighbours is suffering a 50 percent handover failurerate, for example. The optimizing engineer must always look for a practical solutionthat acknowledges the real constraints. For example, in a site suffering very highblocking, it may not be possible to install a second radio carrier (existing cabinets full,lack of space for more, perhaps) and another solution must be found (maybe a newmicro cell and use of a Hierarchical Cell Structure (HCS) perhaps).

It is also important to look for the simplest solutions first. For example, downtilting anantenna to modify coverage before considering a complete change of antenna typeor complicated and risky parameter changes.

Another complicative factor can be the use of Radio Network Subsystem (RNS)equipment from a number of different vendors within a single network. This cancause compatibility problems as not all vendors offer the same features and facilities.

Adjustment of cell parameters is not a precise science. Some trial and error is oftenrequired. It is important to adjust only the minimum number of parameterssimultaneously (one at a time if possible) in order to determine the parameter producing the changes (desirable or otherwise). Parameter changes can beimplemented locally or from the Operations and Maintenance Centre (OMC).

In all optimization activity, it is important to consider possible knock-on effects beforetaking action. Reorienting an antenna could solve coverage problems but causeserious interference problems elsewhere. It is important to consult others, discussthe issues, and perhaps consider alternatives before selecting the final solution.Equally, the appropriate company procedures must be followed when implementing

changes.

Finally, timing is important. Busy hour is not the best time for potentially service-affecting changes of parameters, features, etc. It is necessary to choose the timecarefully and ensure all procedures are followed.

6 SUMMARY OF OPTIMIZATION STRATEGIES





antenna adjustment

omni to sector transmit

parameter tuning

new cells

additional radio carriers

channel types/configurations

deployment of features

customer

quality of service

practical solutions within

constraints

simplest solution first

knock-on effects

consider alternatives

multi-vendor issues

company procedures

timing

Key Optimization Options

Prioritize Activity

Select the Solution

customer

quality of service

reassess if required

Implement the Solution

monitor results

Figure 10

Selecting and Implementing Optimization Solutions

SC2804/S1/v1.1









SECTION 2

OPTIMIZATION SOFTWARE TOOLS

i









1 Software Tools for Optimization 2.11.1 Introduction 2.1

2 Planning and Simulation Tools 2.32.1 Planning Tool Capabilities 2.32.2 The Graphical Display 2.5

2.3 Monte Carlo Simulation 2.112.4 Dynamic Simulations 2.15

3 Drive Test Tools 2.173.1 CW Testing 2.173.2 Live Network Drive Testing 2.19

4 Network Performance Data 2.214.1 Collection, Storage and Processing of Statistics 2.214.2 Key Statistics 2.23

CONTENTS

iii










• identify a range of different software tools that are applicable to the

optimization process

• describe the desired capabilities of different tool types when used to optimize

a WCDMA radio network

• describe how drive tests, ongoing radio coverage tests and traffic

measurements relate to capacity and network optimization

• describe how simulations can be used to analyze optimization problems and

identify potential solutions

• state the role of the NMC/OMC in providing statistical data of various types

• recognize the need for hardware and software tools in relation to testing and

optimization

• recognize the limitations of tool and simulation capabilities

OBJECTIVES

v





1.1 Introduction

There is a wide range of tools available to the optimizer to assist with theoptimization process. Some of these are the same as those used for the planningprocess, for example planning software or drive test tools. Others are specific to theoptimization role. These include system databases, network statistics analysis tools,dynamic simulation software, protocol analyzers, network simulators and parameter tuning tools.

Figure 1 provides a summary of some of the key software tool types that are utilizedfor optimization. These tools can be very complex when applied to UMTS and it isimportant that the optimizer is familiar with their operation and capabilities. Theoptimizer must be able to interpret fully and correctly output information from the tool.While these tools can be very powerful they also have limitations that must beappreciated and allowed for if the correct significance is to be applied to results.

1 SOFTWARE TOOLS FOR OPTIMIZATION





OMC/NMCKPIs

Build andconfigurationdatabases

Protocolanalyzers

Radio planning tools

Static simulation tools

Dynamic simulation tools

Parameter tuning tools

Node B

Drive test tools

RNC

Figure 1

Optimization Tools

SC2804/S2/v1.1





2.1 Planning Tool Capabilities

Planning tools may be modified versions of Second-Generation (2G) planning toolsor they may be dedicated 3G tools. Many operators have both 2G and 3G networksand it is beneficial if the same tool can show and process information about bothsystems simultaneously.

Planning for GSM is usually a simple process of creating coverage predictions basedon balanced uplink and downlink link budgets. However, for UMTS, radio signalstrength predictions are not sufficient. Even if uplink and downlink link budgets havebeen performed that include allowance for system load, specific simulations arerequired to model the effects of traffic. Realistic mixed offered traffic must besimulated as accurately as possible. Therefore the tool needs to have a facility for modelling a variety of traffic and channel characteristics. These are most commonlybrought together to form a service reliability prediction using a Monte Carlosimulation

The optimizer may also be interested in a number of other radio characteristics. For example, prediction of soft handover areas, pilot pollution, Ec/Io values and activeset sizes are very important when considering optimization solutions.

2 PLANNING AND SIMULATION TOOLS





Radio signalstrength prediction

Monte Carlo Simulations

used to produce service

reliability maps

Traffic Modelling

mixed traffic

channel characteristics

demographics

mobility

user characteristics CDMA Factorssoft handover areas

pilot pollution

Ec/lo

UE transmit power

active set size

Figure 2

Planning Tool Capabilities

SC2804/S2/v1.1






2.2 The Graphical Display

The graphical display in any planning tool will contain both foreground andbackground data. Background data includes things like terrain contours, clutter dataand vector data showing roads and railways. It may also be possible to overlay aerialphotos or maps. The display shown in Figure 3a is typical and is taken from the Atollplanning tool produced by Forsk. The display is currently showing terrain data withclutter and vector data on top.

Foreground data includes an indication of site positions, typically with graphical andtext annotations giving an indication of site configuration. On top of this the tool willdisplay the results of predictions and simulations.

Figure 3b shows sites displayed with radio signal strength.

Figure 3c shows a mixed traffic Monte Carlo simulation.

Figure 3d shows predicted soft handover areas.




Figure 3b

Sites and Radio Signal Strength

SC2804/S2/v1.1


Figure 3a

Example Graphical Display



2.7 © Wray Castle Limited SC2804/S2/v1.1




2.8© Wray Castle LimitedSC2804/S2/v1.1


Figure 3d

Soft Handover Prediction

Figure 3c

Mixed Traffic Monte Carlo Simulation




2.3 Monte Carlo Simulation

The Monte Carlo simulation is a critical process in the planning and optimization of UMTS networks. It is not an ideal simulation type in that it is static, but it is a goodcompromise that gives the optimizer a fairly quick and relatively realistic view of likelynetwork operation. It is particularly useful for the optimizer to test the probable impactof a proposed optimization change.

To simulate network operation it is necessary to account for the effects of interference between users in both the uplink and downlink directions. It is alsonecessary to model the effects of power control and mixed traffic. To do this, theMonte Carlo simulation creates a series of snapshots (or drops). For each of thesesnapshots users are randomly scattered over the ground area with weightings for expected traffic density. The tool then uses defined radio parameters to estimatetransmitted power, soft handover requirements and, ultimately, call success rate. Anumber of snapshots can then be combined to produce a statistical analysis of theprobability of coverage for various service types.

2.3.1 Monte Carlo Simulation Inputs

Figure 4 shows some of the most significant input parameters that are required

before a Monte Carlo simulation can be performed. Tools vary in the way trafficprofiles are entered, but typically traffic layers are built up by mapping services touser types and then user types to geographical areas. The result is a map showingthe combined requirement for different services across the map area.

Numerous radio parameters may be required. Many are related to site configurationand radio transceiver performance capabilities. However, some parameters may beadjusted through the optimization process.





Service A

Service B

Service C

Service n

User type A

User type B

User type C

User type n

Area type A

Area type B

Area type C

Area type n

Bit rateRequired Eb/No

Activity factor

PS/CSChannel type

Terminal typeService profileService usageMobility

User types

User density

Monte CarloSimulation

Site details (antenna height, gain, position, etc)Path lossTotal transmit power Pilot power weightingCommon channel power weightingsNoise rise limitEc/Io limitSoft handover thresholdsMaximum active set sizePower control step size

Orthogonality factor

General Radio Parameters

Figure 4

Monte Carlo Simulation Inputs

SC2804/S2/v1.1





2.3.2 Monte Carlo Simulation Output

The output of a snapshot produced through the Monte Carlo simulation will be anindication on the map of user distribution, requested services and connectionsuccess or failure. The example in Figure 5 shows a snapshot based on a simulatedsystem supporting three different user types, each with access to the services listedin the displayed legend. The tool can provide specific data indicating the uplink anddownlink channel performance for each user instance, as shown. Similar collectivestatistics can be produced for site performance.

It is then possible to combine the outputs of a number of snapshots to produce astatistical map for each service type and user type combination.





Figure 5

Monte Carlo Simulation Snapshot

SC2804/S2/v1.1





2.4 Dynamic Simulations

An advantage of static simulations is that they are quick to perform and the resultsare quite easy to interpret. Nevertheless, their accuracy is limited. When staticallysimulated, a call is either active or not, it is either in soft handover or not and power control is stabilized. In a real system there is a lag between measurements andcontrol activity for power control and handover control. Similarly, open loop power control for Physical Random Access Channel (PRACH) establishment and signallingwill precede all call attempts; even those that are unsuccessful. These can beallowed for to some extent in static simulations by including error variables, for example by adding a random error to required transmit power levels, but the mostaccurate results are produced with dynamic simulations.

Dynamic simulations use specialized software that model user activity andmovement over a continuous time frame. This enables much more detailed analysisof network behaviour with a specific set of parameter and configuration settings. Thismethod is more time consuming but is of great value to the optimizer, especially inareas that are sensitive to small changes in settings. This method may also be usedto generate correction factors that will improve the accuracy of results produced instatic simulations.

Care should be taken when setting up dynamic simulations to ensure that they have

a clear objective goal. The results can be difficult to interpret if too many changes insettings are made.





DCH activityincluding closed

loop power control

DCH activity includingclosed loop power control

and soft handover

DCH activityincluding closed loop

power control

RACH activityincluding open loop

power control

UE inactive

Trajectory of UEs is modelled following amap vector such as a road.

Figure 6

Dynamic Simulations

SC2804/S2/v1.1





Drive testing often provides a primary source of information for optimizers

investigating recognized performance problems. Drive testing can be used for a widevariety of network optimization functions including network performance assessment,fault analysis and model tuning. Two basic forms of drive testing are commonlyperformed, Carrier Wave (CW) testing and live network testing.

3.1 CW Testing

This involves the use of a calibrated receiver connected to a data storage device,typically a laptop or a Personal Digital Assitant (PDA). The receiver may be capableof measuring more than one frequency simultaneously.

For UMTS it is useful if the receiver is capable of providing measurements of Receive Signal Code Power (RSCP) and Ec/Io for individual cells. However, basicCW testing measuring radio signal strength may be used on individual frequenciesfrom a test transmitter for basic path loss estimation.

CW testing is most commonly used for propagation model tuning and verification.The example in Figure 7 shows an overlay of CW test data on an empiricallygenerated signal strength prediction. These differences can be analyzed to calculatea standard deviation for the cell. This can then be used to modify the ‘k’ values in the

empirical model.

3 DRIVE TEST TOOLS





Figure 7

CW Testing

SC2804/S2/v1.1





3.2 Live Network Drive Testing

This type of drive testing involves the connection of a test mobile (usuallyincorporating test software) to a logging device such as a laptop or PDA. A series of calls are made, either manually or automatically, and all events and signalling duringthe calls are recorded. It is particularly useful to record measurement data from thetest mobile, both during calls and while in idle mode.

UMTS offers the possibility to provide modified measurement commands toindividual mobiles. This would mean that test mobiles could be asked to measure alarger neighbour set and provide more detailed measurements. The range of measurements that can be specified for UMTS is extensive.

The recorded data captured during a drive test is then replayed using a drive testanalysis tool. This may be a specialized tool, but many planning tools will alsooverlay some drive test data. Drive test analysis tools will typically use recordedpositional information to provide a rolling map display for real-time or slow-timereplay of drive test logs. Many analysis tools provide a protocol analysis function sothat signalling can be decoded. This is particularly useful when analyzing thereasons for call failure.

Figure 8 shows part of a drive test log overlaid on a graphical display in a planning

tool.





Figure 8

Live Network Drive Testing

SC2804/S2/v1.1





In a UMTS network, performance data is available in the form of raw statistics from

all major network elements in the core network such as Mobile-services SwitchingCentres (MSC), messaging platforms, databases and other service platforms.Similarly, performance data can be gathered form all network elements in the UMTSTerrestrial Radio Access Network (UTRAN) such as RNCs, Node Bs, transmissionnodes and Location Management Units (LMU). These statistics are essential for theday-to-day operation of the network, providing data for ongoing performanceevaluation against targets. This information is also critical for the optimizer because itmay be used for problem analysis and provides a means of assessing the successor otherwise of optimization solutions.

4.1 Collection, Storage and Processing of Statistics

All network elements, for example an RNC, collect and store statistical data locally.These raw statistics, of which there are many different types, are uploaded to theOMC/NMC at regular intervals. Usually they can also be read locally using a laptop.The uploads are carried out using Operations and Maintenance (O&M) data links,normally utilizing part of the transmission infrastructure. The upload interval could beas short as every five minutes, but is more likely to be every 15 or even every 30minutes. It is possible for the most important statistics to be uploaded morefrequently than other data in some systems.

Raw statistics are sometimes called counters. The raw statistics can be viewed astabular or graphical data, or further processed to provide key statistics, which arealso known as metrics or Key Performance Indicators (KPI).

4 NETWORK PERFORMANCE DATA





Storage inrelationaldatabase

Reporting tool

Data collectionprocess

Statisticalreports (KPIs)

Localaccessto data

OMC/NMC

GraphicalTabular

Figure 9

Gathering Network Performance Data

SC2804/S2/v1.1





4.2 Key Statistics

Key statistics are the KPIs that are used to judge whether the network is working toits design criteria. They are created through the processing of raw statistics. For example, raw statistics may be uploaded from each cell regarding the number of callrequests, the number of successful attempts and the number of unsuccessfulattempts. These would all be provided for a defined measurement period, perhapsevery 15 minutes. If all these results are summed for all the cells on an RNC over a24-hour period, then a KPI could be produced representing average call successrate for each day. Typically this would be divided into success rates for eachdefinable call type, for example voice, video telephony, low-rate packet data andhigh-rate packet data.

KPIs will be required for many different aspects of the operational network’sperformance. Figure 10 provides some examples of things that may be included, butit is up to individual operators to determine the most appropriate KPIs.

KPIs falling below an expected threshold may trigger optimization activity. Thesestatistics in themselves may be useful for the optimizer, but more detailed analysis isoften required to isolate a problem. For example, the call success rate mentionedabove may be studied on an hourly basis in order to identify a time period when theproblem occurs.

More detailed analysis may also be set up when a new feature is introduced on atrial basis. Because of the potentially very large amount of data generated, it isbeneficial if particular information about performance is targeted for detailed analysisin respect of the new feature. However, standard statistics should also be monitoredin case the feature has an unexpected effect.





number of dropped calls

number of soft handovers

number of hard handovers

handover success rate

average percentage of calls in soft

handover average transmit power (uplink and

downlink) by call type

cell throughput

RNC throughput

QoS statistics for packet data

average call hold time

average mobility of users per cell

average range of users in a cell

Connected Mode Related

success rate

location update

routing area update

UTRAN registration area update

Idle Mode Related

total attempts

location update

routing area update

UTRAN registration area update

paging success rate

RACH success rate

successful channel allocations

successful PDP context activations

average duration for call establishment

average range from which call attempts are

made

Set-up Related

Figure 10

Typical Key Statistics

SC2804/S2/v1.1









SECTION 3

OPTIMIZING COVERAGE AND

CAPACITY

i









1 Link Budgets 3.11.1 Load Factor 3.31.2 Load Factor and Noise Rise 3.51.3 Optimization Considerations for Load Factor 3.71.4 Mixed Traffic and Load Factor 3.9

2 Coverage and Capacity Optimization Issues 3.112.1 Coverage Solutions 3.112.2 Capacity Solutions 3.152.3 Adaptive Voice Channels 3.272.4 Secondary Scrambling Codes 3.31

3 Traffic Scenarios 3.333.1 Introduction 3.333.2 Uplink Limited Systems 3.333.3 Downlink Limited Systems 3.33

4 Evolving Radio Access Architecture 3.354.1 Rollout Architecture 3.354.2 Antenna Azimuths and Beamwidth 3.374.3 More Sectors or More Cells? 3.394.4 Use of Repeaters 3.414.5 Basic Considerations for Indoor Coverage 3.51

5 Exercise 1 – Urban Capacity and Coverage 3.53

6 Location Services (LCS) 3.576.1 Introduction 3.576.2 Quality of Service 3.596.3 Factors Affecting Accuracy of Location Information 3.616.4 Response Time 3.636.5 Cell ID Based Positioning Mechanism 3.656.6 Observed Time Difference Of Arrival (OTDOA) 3.696.7 Network-Assisted Global Positioning System (GPS) 3.79

7 Propagation Modelling 3.817.1 Empirical Models 3.817.2 Deterministic Models 3.85

7.3 Comparing Models and Their Effects 3.87

CONTENTS

iii










• perform link budget calculations to verify cell size and traffic load capabilities

in mixed traffic scenarios

• describe the impact of coverage and capacity expected for a range of mixed

traffic scenarios

• describe the conditions in which a cell may become uplink or downlink

limited

• describe conditions in which a system may be coverage or interference

limited

• describe how the rollout architecture for a UMTS network can be evolved to

expand capacity and coverage

• discuss the merits of cell splitting and multicell sites

• discuss the merits of using repeaters to improve coverage

• describe how masthead amplifiers can be used to improve coverage and

capacity in a UMTS system

• identify suitable propagation models and explain the need for accurate model

tuning

• state the requirements for optimization of location capabilities in the radio

access network

OBJECTIVES

v





A link budget must be performed in both the uplink and downlink directions. For GSM

this only involves radio factors such as transmit power, receiver sensitivity, feeder losses and antenna gains. The aim is to find a maximum path loss that is acceptablein both the uplink and the downlink directions. For GSM, the result of this calculationis static since it is not altered by cell load.

In UMTS the link budget is not static because it is affected by cell load. There are tworelated aspects to this: the fact that the technology is CDMA-based and also theneed to support mixed traffic.

In a link budget for a CDMA-based system, account must be taken of theinterference present due to other users. This is a factor of serving cell load and also,to a lesser extent, of neighbour cell load. The resulting interference level is known asnoise rise. It is necessary to allow a margin for noise rise when calculating the linkbudget. This margin is referred to as the interference margin.

The noise rise is calculated from the load factor of a cell. The value of load factor islargely dependent on two factors: the channel processing gain and the required valueof Eb/No at the receiver output. Both these factors will be different for differentservices with different QoS requirements. Thus a realistic value of load factor canonly be achieved if realistic mixed traffic cases are considered. An importantconsideration for the optimizer will be the degree of correlation between the

estimated traffic load used at the planning stage and the real traffic load when anoptimization problem arises.

1 LINK BUDGETS





Node B UE

Radio Parameters

Maximum acceptable

path loss

Downlink link budget

Uplink link budget

Interference Margin

Noise Rise

Load Factor

Mixed Traffic

Figure 1

Link Budget Inputs

SC2804/S3/v1.1





1.1 Load Factor

In practice the received signal power at a cell and at the UE contains both wantedchannel data and unwanted interference. The theoretical maximum load on a cellwould be when all the received power was wanted channel data. The load factor isthe ratio of wanted power to unwanted power and is a measure of how close a cell isoperating in relation to its theoretical maximum load.

The calculation of uplink and downlink load factors differs slightly because of therelative positions of the transmitters and receivers. In the uplink direction thechannels are transmitted from different locations, but are all received in the samelocation. This means that the effect of neighbour cells can be considered constant for all channels. In the downlink direction all channels are transmitted from the samelocation but received in different locations. The effect of neighbour cell interferencevaries as a result of the UE’s location and, ideally, this should be included in the loadfactor calculation. Additionally, a factor must be also allowed in the downlink toaccount for lack of orthogonality between variable-length codes in a multipathchannel.

Figure 2 provides expressions for uplink and downlink load factor calculation. Notethat these expressions do not allow for a mixed traffic case as shown. However, thiscould be accounted for simply by summing the load factor estimate for each

individual traffic type.





=+

+=

ν

η

=+

= +

ν

−αη

ηUL = UL load factor

= activity factor for UE j j

= neighbour cell interference factor j

= orthogonality factor j

= bit rate for UE jR j

= noise spectral densityNo

= energy per bitEb

= chip rateW

= an individual UE j

= number of UEs in the cell NN

ηDL = DL load factor

Figure 2

Uplink and Downlink Load Factors

SC2804/S3/v1.1





1.2 Load Factor and Noise Rise

Noise rise is derived from both the uplink and the downlink load factors (η) in thefollowing way:

Noise rise = 11 – η

Noise rise is more usefully expressed in decibels for inclusion in the link budget asan interference margin; in which case the expression becomes:

Noise rise (dB) = –10log10

(1 – η)

Figure 3a shows the relationship between load factor and noise rise expressed indecibels. It can be seen that noise rise tends to infinity as load factor approaches100%. It is not advisable to plan a system with very high load factors. The shape of the curve indicates that at high load factors small changes in load give rise todramatic changes in noise rise. A system planned to carry such loads would requirean impossibly high interference margin or it would suffer extreme cell breathingeffects. This is perhaps most graphically illustrated when looking at a linear representation of the curve as shown in Figure 3b.

If the maximum load factor is planned to be in the region of 60% to 80% then thecurve is relatively flat. A system planned in this way requires a more manageableinterference margin leading to achievable link budgets. In addition, it should showminimal cell breathing up to the intended cell capacity limits.





50

40

30

20

10

00% 20% 40% 60% 80% 100%

NoiseRise

)(r otcaFdaoL

Figure 3b

Load Factor and Noise Rise (Linear)

SC2804/S3/v1.1


18

16

14

12

10

8

6

4

2

00% 20% 40% 60% 80% 100%

NoiseRise (dB)

)(r otcaFdaoL

Figure 3a

Load Factor and Noise Rise (Logarithmic)




1.3 Optimization Considerations for Load Factor

At rollout a UMTS network will have a relatively small number of subscribers, whoare not likely to make full use of high-rate data services. The operator’s aim at thisstage will be to maximize coverage. Capacity in the network is unlikely to be aproblem. Therefore it makes sense to select a fairly low load factor as a basis for coverage planning with macro cells only.

Consider Figure 4. A load factor of 50% gives rise to a 3 dB noise rise. Including thisas the interference margin in the link budget places a small, but still significant,limitation of maximum acceptable path loss. For example, if the operator wished toprovide contiguous coverage in an urban area offering at least 144 kbit/s to class 3UEs, a typical link budget might suggest a maximum acceptable path loss of about145 dB. This can be interpreted in terms of cell radius using, for example, theCOST231-Hata model. When not considering the interference margin this gives acell radius of about 1.2 km. When allowing for a 50% load factor it is necessary toadd another 3 dB interference margin. This reduces cell radius to 1 km.

The 50% load factor would be enforced by the Call Admission Control (CAC) policyin the RNC. As traffic levels rise in the network, the cell load factor limit will beginblocking calls with a resulting fall in grade of service. Simply increasing the permittedload factor to alleviate this is not a sensible solution. For example, if the CAC policy

was modified to allow a load factor of 75%, then noise rise would be increased to 6dB. When factored into the link budget as interference margin cell radius is reducedto approximately 800 m at busy times. This could leave coverage gaps in thenetwork.

This could be dealt with by the introduction of either in-fill cells or a hierarchicalcellular architecture incorporating micro cells. Micro cells used simply to absorbtraffic load rather than provide extended coverage could be planned on theassumption of high load factors.

Typical load factor figures for macro cells would be in the range 50% to 60%. This

gives a good compromise between maximizing coverage potential and maintaining areasonable traffic load. Micro cells are added with less emphasis on cell radius andmore emphasis on capacity. Typical load factors for micro cells could be in the regionof 75% to 80%.





6

3

50% 75% 100%

NoiseRise (dB)

)(r otcaFdaoL

ataH-132TSOCgnisU

Bd541ssolhtap.xaMEU3ssalc,aer anabr u,s/tibk441

:elpmaxE

mk2.1suidar lleCBd0=esir esioN%0=r otcaf daoL

mk1suidar lleCBd3=esir esioN%05=r otcaf daoL

m008suidar lleC Bd6=esir esioN

%57=r otcaf daoL

Figure 4

Load Factor Illustration

SC2804/S3/v1.1





1.4 Mixed Traffic and Load Factor

It can be seen from the load factor equations that Eb/No requirements for a particular service contribute to determining the load factor. The output Eb/No requirement itself is dependent on the service type and the error protection being applied in thechannel. For example, a typically allowed figure for a standard voice service wouldbe 5.5 dB, whereas high-rate data is often taken to be much lower, perhaps as lowas 1.5 dB or even 1 dB. The reason for this low Eb/No figure is the assumed use of more powerful error protection schemes such as Turbo coding and the relaxed delayconstraints permitting retransmission. A lower Eb/No figure means that total cellthroughput can be higher for a given load factor. The mobility and the geographicallocation of the UE may also influence Eb/No requirements because the prevailingchannel conditions will have an impact on error characteristics.

Another important factor is the level of neighbour-cell interference contribution. Thisis usually assumed to be higher in macro cells than in micro cells. This is becausemicro cells tend to be sheltered by street canyons and therefore suffer less fromneighbour-cell interference. Again, a lower interference factor means more cellthroughput for a given load factor.

Finally, it can also be assumed that a higher load factor can be tolerated on microcells than on macro cells because coverage and ultimate cell range is less of a

concern. The UE is likely to be much closer to a micro cell and therefore a larger interference margin can be included in the link budget.

Figures 5a and 5b show calculations of cell throughput in kbit/s for different cell typesand service types. Calculations have been performed for the macro cell with 50%and 60% load factors, and for the micro cell with 75% and 80% load factors. Thisillustrates the extremes of variation that are to be expected in usable cell capacity for UMTS cells. These calculations assume that all users in each scenario will be usingthe same service type. In reality, a cell could be expected to deal with a dynamic mixof service types, in which case the throughput will be some amalgam of the valuesshown here.






ServiceBit

Rate(kbit/s)

ActivityFactor

Eb/No(dB)

LoadFactor

N-cellInterference

Factor

Number of

Channels

Total CellThroughput

(kbit/s)

Voice 12.2 0.6 5.5 75% 1.1 101 1232.2

Low packetdata

64 0.9 2.5 75% 1.1 27 1664

Mediumpacket data

144 0.9 1.5 75% 1.1 15 2160

High packetdata 384 0.9 1 75% 1.1 6.7 2572.8

Voice 12.2 0.6 5.5 80% 1.1 108 1317.6

Low packetdata

64 0.9 2.5 80% 1.1 28 1792

Mediumpacket data

144 0.9 1.5 80% 1.1 16 2304

High packetdata

384 0.9 1 80% 1.1 7.1 2726.4

Figure 5b

Micro Cell – Mixed Traffic

ServiceBit

Rate(kbit/s)

ActivityFactor

Eb/No(dB)

LoadFactor

N-cellInterference

Factor

Number of

Channels


(kbit/s)

Voice 12.2 0.6 5.5 50% 1.3 57 695.4

Low packetdata

64 0.9 2.5 50% 1.3 14 896

Mediumpacket data

144 0.9 1.5 50% 1.3 8 1152

High packetdata

384 0.9 1 50% 1.3 3.7 1420.8

Voice 12.2 0.6 5.5 60% 1.3 68 829.6

Low packet

data

64 0.9 2.5 60% 1.3 18 1088

Mediumpacket data

144 0.9 1.5 60% 1.3 10 1440

High packetdata

384 0.9 1 60% 1.3 4.5 1728

Figure 5a

Macro Cell – Mixed Traffic




Coverage and capacity are closely linked in UMTS; nevertheless, it is possible to

consider independent optimization strategies for each characteristic. In some casesbenefits arising from successful optimization activity may result in improvements toboth coverage and capacity, but even here it is possible to weight the effect toinfluence one or the other more noticeably.

2.1 Coverage Solutions

Coverage is likely to be of prime concern when a network is in the rollout phase. Themain limiting factor will be the low transmit powers from a UE, most UEs being class4 with a maximum output power of 21 dBm (0.125 W). This, coupled with anoperating frequency in the region of 2 GHz, means a restricted uplink power budget.Well-established radio techniques and some CDMA-specific techniques can be usedto improve coverage. These include:

• antenna height

• antenna gain/types

• antenna alignments

• low noise amplifiers

• repeaters• soft handover gain

2.1.1 Antenna Solutions

Rollout Node Bs will be predominantly macro cells with antennas mounted relativelyhigh compared to average building height. When such cells are used to maximizecoverage they will probably be unbalanced such that the potential downlink radius issignificantly greater than the uplink radius. This means that different antenna gains

need to be used to balance the link. A common approach is to use omni transmit andsector receive over three sectors. Optimization attention will be focused on uplinkantenna types, gains and alignments to maximize coverage and minimizeinterference. Close attention should be paid to simulation of performance effectscaused by antenna installation errors that are within the tolerances set for site buildand acceptance. It may also be worth considering higher gain antennas, perhapswith more than three sectors.

2 COVERAGE AND CAPACITY OPTIMIZATION ISSUES





Node B UE

Typically class 421 dBm (0.125 W)

Weak uplinklink budget

Unbalanced downlinklink budget

Antenna Coverage Improvements

antenna configuration (omni transmit)

antenna type

antenna gainantenna alignment

build tolerances

Figure 6

Antenna Coverage Solutions

SC2804/S3/v1.1





2.1.2 Low Noise Amplifiers (LNA)

The use of Low Noise Amplifiers (LNA) is a well established technique for boostinguplink power budget performance. These are sometimes referred to as Mast Head Amplifiers (MHA) or Tower Mounted Amplifiers (TMA). They reduce the noise figureat the input to the receiver, which helps to compensate for the low UE transmitpower.

The reduction in noise floor created by an LNA could also be used to increasecapacity because it allows for more noise rise.

2.1.3 Repeaters

Repeaters may be used for coverage improvement in areas that are not likely topresent high traffic loads. They should not be used where it is predicted that trafficload will increase significantly over time unless the site can be upgraded to a Node Bwith ease. If planned with care a repeater may also provide some increase incapacity.

2.1.4 Soft Handover Gain

While in soft handover the UE is benefiting from uplink and downlink spatial diversityin the link. This produces a gain usually referred to as soft handover gain. Softhandovers reduce overall capacity in a network because a call requires multiplechannel resources. However, in areas where coverage is of prime concern it may bepossible to reduce handover margins to increase the soft handover area.





Low Noise Amplifiers (LNA)

reduce receiver noise floor

boost uplink link budget

increase capacity

Repeaters

low traffic or rural areas

in-building coverage

cheaper than new Node B

could provide some capacitybenefits

Weak uplinklink budget

UETypically class 4

21 dBm (0.125 W)

Node B

Figure 7

Low Noise Amplifiers and Repeaters

SC2804/S3/v1.1





2.2 Capacity Solutions

As a network matures the customer base will increase, as will the range of servicesoffered to subscribers. An early and very important function for optimization teamswill be to evolve the radio access network from a coverage-oriented design towardsa capacity-oriented design. This will involve a mixture of architectural changes andthe introduction of new features as they become cost effective. This will include:

• use of more frequencies

• use of UMTS Time Division Duplex (TDD) mode

• in-fill cells• Hierarchical Cell Structures (HCS)

• indoor coverage solutions

• more sophisticated 2G interworking

• antenna configuration changes

• antenna orientation/downtilt

• adaptive voice channels

• secondary scrambling codes

• Multi-User Detection (MUD)

• transmit diversity

2.2.1 More Spectrum

Most UMTS operators have licences for enough spectrum to operate more than oneFDD carrier pair. Typically an operator may be able to implement two or three carrier pairs. These could be used in a variety of ways, but essentially an operator maychoose to use then as independent cell layers or to provide more capacity within a

cell layer. Overall the highest capacity will probably be achieved through the use of hierarchical cell structures partitioned by frequency. It is important for optimizers tobear in mind that different solutions may suit different locations and an operator canuse different strategies in different geographical regions if appropriate.

Even where UMTS operators have only one Frequency Division Duplex (FDD)carrier pair there may still be scope for spectrum sharing. This option would increasecapacity and reduce infrastructure costs for the operators and is therefore worthy of consideration.





Progressivelyintroduced as

second carrier onmacro cell sites or

for a micro celllayer

Used atrollout on

macro cells

Progressively

introduced as amicro cell or

pico cell layer

5 MHz

Example UMTS Licence

Progressively


pico cell layer

5 MHz 5 MHz 5 MHz

FDD (x3 pairs)TDD (x1)

Figure 8

Additional Radio Carriers

SC2804/S3/v1.1





2.2.2 UMTS TDD Mode

Many UMTS operators have licences that include spectrum for TDD mode radiocarriers. Typically this will be a single carrier, but TDD mode is a very flexibletechnology solution. Although it is a UMTS technology the optimizer will need to treatit as a different radio access technology and integrate it as such.

Potentially the cell sizes for a TDD mode cell and an FDD mode cell are the same;however, the TDD technology is more suited to non-symmetric data applications.This makes TDD mode a candidate technology for the implementation of pico cellsand indoor coverage solutions.

It may also be the preferable technology solution for special project cell where, for example, it may be desirable to stream high-quality video.





Progressivelyintroduced as

second carrier onmacro cell sites or

for a micro celllayer

Used atrollout on

macro cells

Progressively


pico cell layer

5 MHz

Example UMTS Licence

Progressively


pico cell layer

5 MHz 5 MHz 5 MHz

FDD (x3 pairs)TDD (x1)

Figure 8 (repeated)

Additional Radio Carriers

SC2804/S3/v1.1





2.2.3 In-Fill Cells and HCS

Ultimately, the need for more capacity will always lead to a need for more cells. Thefirst step in this process may be to in-fill new cells between the macro cells of arollout architecture. This will need considerable attention from the optimization teamto target new capacity appropriately and minimize the potential negative impact onexisting cells.

As a network develops the new cells may be implemented as overlays on existingcoverage. In this case parameters and procedures required for the effectiveoperation of HCS will need to be introduced. These should be monitored and tunedby the optimization team.





Coverage areas of rollout cellsreduced with downtilt and pilotpower reductions.

Optimization needed to ensure

new in-fill is beneficial notdetrimental to networkperformance.

RolloutNode B

Rollout

Node B

RolloutNode B

New in-fillNode B

Figure 9

In-Fill Cells and HCS

SC2804/S3/v1.1





2.2.4 Interworking with 2G

Most operators are overlaying a new UMTS network onto a mature and (usually)well-optimized GSM/General Packet Radio Service (GPRS) 2G network. UMTSoffers capabilities above a 2G network, but many of the offered services can becarried adequately on a 2G network; for example, voice or messaging services.Therefore, balancing the traffic load in the most appropriate way between the 2Ginfrastructure and the 3G infrastructure is an important optimization task. The mainmechanism for this will be effective and appropriate settings of triggers for inter-RAThandovers, but it may also impact on admission control.





Split trafficaccording to QoS

requirements

3G 2G

Optimize handovers and

reselections to account forQoS requirements

Figure 10

Interworking with 2G

SC2804/S3/v1.1





2.2.5 Antenna Solutions

Several aspects to antenna optimization can influence capacity in a UMTS network.Firstly, antenna type, orientation and downtilt will need changing as more cells areadded in the system. As in-fill cells are added considerable reorientation may berequired; typically, a 30° azimuth change is applied in an area where cells are placedin an even hexagonal pattern. In theory this maintains an even geographical areasplit between original and new cells. However, in more realistic and variableenvironments the exact orientation changes need to be chosen for best performance.It is generally to be expected that increasing amounts of downtilt will be applied astraffic load and the number of cells increase. In some cases reduction in antennaheights may be considered appropriate.

Cells may also be changed from omni to sector transmit, which may require theaddition of new antennas or simply the rerouting of feeder runs and the addition of new power amplifiers.

One option for capacity increase would be the introduction of cells with more thanthree sectors. Again, this will require new antenna and feeder runs at the site. Theextra equipment required at a site may also mean that more space and mastreinforcement are required. The cost of this will be a factor determining whether thisapproach is used or not.

Finally, there are more advanced antenna types that could be used to increasecapacity. These may initially take the form of beam-forming antenna arrays, followedby dynamically adaptive beamforming arrays (sometimes called smart antennas).





Omni transmit tosector transmit

Sites with morethan three sectors

Beam formingantennas

Figure 11

Antenna Solutions for Capacity

SC2804/S3/v1.1





2.2.6 Multi-User Detection (MUD) and Transmit Diversity

Multi-user detection and transmit diversity are two optimal features that could beintroduced to gain extra capacity from an existing cell plan.

Multi-user detection is a form of noise cancellation. It utilizes knowledge about thenature of the interference from one channel onto another to correct for the distortioncaused. It is applicable only to the uplink direction. New software and possibly somenew hardware would be required to upgrade a Node B for this capability.

Transmit diversity utilizes two transmit antennas mounted so as to provide spacediversity. Transmissions are marked such that the UE can identify which antenna aparticular copy of the received signal was transmitted from. The result is that the UEcan optimally combine multiple copies of the received signal with a significantdiversity gain figure. There are several different modes of operation available for transmit diversity, but all would require software upgrades and a significant amountof new hardware adding on the site.





Multi-User Detection (MUD)

uplink channel performance improvementhardware and software changes in Node Bsignificant capacity improvements

Transmit Diversity

downlink channel performance improvementsignificant hardware and software changes inNode B

significant capacity improvements

Figure 12

Multi-User Detection and Transmit Diversity

SC2804/S3/v1.1





2.3 Adaptive Voice Channels

The Adaptive Multi-Rate (AMR) voice codec is designed to utilize variable bit ratesfor achieving an optimal balance between quality and capacity in carried voice traffic.

The AMR codec is applicable to both GSM and UMTS. It provides eight voice codingrates set between 4.75 kbit/s and 12.2 kbit/s. In addition, it supports discontinuoustransmission with Silence Descriptor (SID) frames at an effective rate of 1.8 kbit/s.The voice coding rates are designed to be adjusted dynamically according to radioconditions. In theory the rate can be changed every 20 ms, but air interface delaysand processing time mean that in practice the adjustment rate will be lower than this.

The voice-coding rate will be lowered, and error protection overhead increased, aschannel conditions worsen. This results in a more robust channel exhibiting moreconsistent voice quality. Using AMR in this way can provide an overall improvementin perceived voice quality for users and an increase in capacity for the operator.

There are several ways in which the benefits of the AMR coder can be applied, andits operation in GSM and UMTS is slightly different. For UMTS the eight bit rates areused to provide a smooth trade off between voice quality and capacity in the networkas a whole. Additionally, the 7.4 kbit/s codec mode provides compatibility with legacynetworks in North America, and the 6.7 kbit/s codec mode provides compatibility with

those in Japan.

For GSM, coarse adjustment is provided through the selection of a full-rate or a half-rate channel mode. These two channel modes relate to a gross channel bit rate of either 22.8 kbit/s for the full-rate channel mode, or 11.4 kbit/s for the half-ratechannel mode. The half-rate mode for AMR (7.95 kbit/s) provides significantlyimproved voice quality when compared to the standard GSM half-rate codec. In boththe full-rate and half-rate channel modes the codec mode may then be changeddynamically between the defined voice coding rates according to channel conditions.This provides a very flexible tool enabling operators to achieve an effective balancebetween capacity and quality.





Codec Modes

UMTS Operation and

GSM FR Channel Mode(kbit/s)

GSM HR Channel

Mode(kbit/s)

AMR_12.20 12.2 (GSM EFR)

AMR_10.20 10.2

AMR_7.95 7.95 7.95

AMR_7.4 7.4 (IS-136 EFR) 7.4 (IS-136 EFR)

AMR_6.7 6.7 (PDC EFR) 6.7 (PDC EFR)

AMR_5.9 5.9 5.9

AMR_5.15 5.15 5.15

AMR_4.75 4.75 4.75

AMR_SID 1.8 1.8

Figure 13

AMR Codec Modes and Application

SC2804/S3/v1.1





2.3.1 Adaptive Voice Channel Benefits

The AMR voice coder enables the operator to change the characteristics of the voicechannel for more capacity (or more coverage) by accepting lower voice codingquality. Figure 14 shows how the net rate of the voice channel, which representscoded voice plus error protection overhead, remains constant. The channel adapts tochanging quality by varying the ratio of coded voice and error protection overhead.

With more error protection overhead the required Eb/No is reduced and theprocessing gain is increased. Figure 14 also shows the effect this may have on cellcapacity for a macro cell supporting only voice calls.





ServiceBit

Rate(kbit/s)

ActivityFactor

Eb/No(dB)

LoadFactor

N-CellInterference

Factor

Numberof

Channels


(kbit/s)

AMR_12.20 12.2 0.6 5.5 60% 1.3 68 829.6

AMR_7.95 7.95 0.6 5.0 60% 1.3 118 938.1

AMR_5.9 5.9 0.6 4.5 60% 1.3 178 1050.2

Channel

quality

Error protection

Coded voiceaccording to

selected codecmode

Net channel rate

Figure 14

Adaptive Voice Channel Benefits

SC2804/S3/v1.1





2.4 Secondary Scrambling Codes

Separation of channels in the downlink direction is facilitated by the OVSF codes.These are designed to be fully orthogonal within their set providing that the rules for allocation of codes from the code tree are followed.

These rules are illustrated in Figure 15a. Once a code has been allocated, no codederived from it or from which it is derived to the root of the tree may be used. Anefficient code allocation algorithm is required to make best use of the availablecodes. Potentially most problematic is the allocation of high spreading factor codes.In this case the allocation of one code to a low-bit-rate user removes a substantialportion of the tree. The most efficient allocation strategy is to allocate high spreadingfactor codes from the same or from related branches if possible. However, even withan efficient code allocation algorithm there may be no available codes even when thecell has not reached its load factor limit.

The result is that a cell may become code limited rather than interference limited.Secondary scrambling codes can be used to overcome this problem. There are 15secondary scrambling codes associated with every primary scrambling code. Theallocation of secondary scrambling codes provides additional partitioning between acell’s downlink channels. They may be used on any of the downlink channels other than the Common Pilot Channel (CPICH) and the Primary Common Control Physical

Channel (PCCPCH). This means that capacity will not be limited by the possibility of running out of codes.

Additionally the use of multiple code lengths simultaneously in a multipath channelleads to a reduction in orthogonality. This is usually allowed for by an extrainterference factor in the load factor calculation. Secondary scrambling codes couldbe used to partition different channels using different spreading factors and thusreduce intra cell interference, ultimately increasing capacity.






Primary Scrambling Code

Secondary Scrambling Code

OVSF codes can bereused when covered

by a secondaryscrambling code

Figure 15b

Secondary Scrambling Codes

Root

Allocated code

Non-allocatable codes

To the topof the tree

Figure 15a

Secondary Scrambling Codes




3.1 Introduction

Second-generation systems can be broadly categorized as being either uplink or downlink limited. Once categorized as one or the other they can be treated as suchirrespective of traffic load. The determining factors are radio related, and typicallymost second-generation technologies, particularly those operating at the higher endof UHF, are uplink limited. System balance is usually achieved by adjusting gainsand losses in the antenna path.

UMTS is also affected by radio factors and an unloaded cell is generally assumed tobe uplink limited. This results mainly from the high transmit frequency and the lowUE transmit power capability.

3.2 Uplink Limited Systems

The relationship between load and coverage in the uplink direction is fairlystraightforward. An increase in load is accompanied by an increase in noise rise.This is accounted for by including an interference margin in the link budget. Theresult is a radius that reduces as load increases; the familiar cell breathing effect.

Figure 16 is a generalized graph showing a typical relationship between the

maximum acceptable path loss in decibels and the cell load in kbit/s. It can be seenthat in this example the cell is uplink limited when load is below about 650 kbit/s.This is likely to be the case for most cells in the rollout phase of a UMTS network.

3.3 Downlink Limited Systems

In the downlink direction noise rise also increases with traffic load. However, therelationship between noise rise and the maximum acceptable path loss is further complicated by the limitations of the cell’s power amplifier. There will always be afinite amount of power available, which must be divided between all the downlink

channels. As the number of channels increases with cell load, so the amount of transmit power available per channel in the downlink decreases. Thus theinterference margin and cell transmit power become load-dependent variables in thelink budget. The result is that at high cell loads the cell becomes downlink limited.

This effect is further accentuated when higher utilization of packet data servicesbecomes more common. Many of these services are typically unbalanced such thatthey place more load in the downlink than the uplink.

3 TRAFFIC SCENARIOS





Uplink

Downlink

Downlink limited

Uplink limited

165

160

155

150

145

140

135

100 200 300 400 500 600 700 800 900 1000 1100

MaximumPath Loss

(dB)

Cell Load(kbit/s)

Figure 16

Coverage and Capacity Limitations

SC2804/S3/v1.1





4.1 Rollout Architecture

Rollout topology can be expected to be wholly, or at least largely, implemented withsingle-carrier macro cells. Ideally these should be placed in an even hexagonalpattern with identical radio configurations. In reality, topological and morphologicalconsiderations will result in localized variations. Additionally, most operators will needto reuse existing sites and integrate an element of site sharing into the plan toreduce build costs. The result will be an irregular pattern of cells with variations inantenna height, type, and alignment as well as transmit power.

Simulations suggest that making small deviations from the regular hexagonal patternto suit the nature of the planned area has very little effect on system performance.Perhaps not unexpectedly, the regular hexagonal pattern turns out not to be theoptimal plan in realistic non-homogeneous areas.

However, if these variations do not suit the terrain and are random in nature thenthey can lead to a deterioration in system performance. There are probably two mainreasons for this. Firstly, a lack of accurate terrain and demographic data in thesimulation tool. Secondly, the need to reuse existing cell sites that will not be ideallysuited to UMTS operation.

4 EVOLVING RADIO ACCESS ARCHITECTURE







4.2 Antenna Azimuths and Beamwidth

As the network is expanded and evolved it may be necessary to realign antennas inorder to optimize traffic distribution between cells. For example if in-fill cells areadded or if hierarchical cells are added without using a second frequency.

The diagram shows one of the considerations for azimuth changes. The anglebetween two adjacent antennas has been reduced. This is likely to have been donein an effort to improve the effectiveness of coverage with planned benefits for linkbudgets and capacity. The result will be that the overlap area between the two cellsis increased. There are two significant consequences of this.

Firstly, it will change the soft handover relationship between the two cells. Thelocation of the soft handover area will move and, depending on handover triggers,antenna type and local topology it may either increase of decrease in size. Inevitablethese changes will have an impact of total load carried by these two cells.

Secondly, the isolation between the two antennas will be reduced. The extent bywhich it is reduced would depend on antenna type and their relative mountingpositions. The reduction in isolation will increase mutual interference in both theuplink and downlink directions; again reducing capacity.

Each case must be considered independently and, if possible, simulations carriedout to identify potential problems. Any simulation should allow for build tolerances.Using antennas with a narrower horizontal beamwidth might be considered if problems are suggested by the simulations.





Soft

handover

area

will change

in locationand size

Isolationreduced

e.g. 30°

Figure 18

Changing Antenna Azimuths

SC2804/S3/v1.1





4.3 More Sectors or More Cells?

When attempting to find more capacity from an existing infrastructure it is alwaysbest to look for the most economic solution. Building sites with more than three cellsin GSM is rare because of high costs for a relatively low return. However, the addedinterference tolerance gained by using more cells on a site in UMTS translatesdirectly to increased capacity. This makes such sites potentially more cost effective.

In some locations adding more cells to a site may be preferable and cheaper thanbuilding new in-fill or hierarchical cells to deal with increased traffic load. However,this will not always be the case. It is important for the optimizer to consider spaceand facilities at a site as well as aesthetics in some sensitive locations. Micro cellsand pico cells have much less visual impact than the large tower required to supporta six-cell site. Possible evolution should also be considered. The more commonapplication of advanced optimal features such as multi-user detection or beamforming antennas may be a better longer-term evolutionary path for a site.Micro cells and pico cells may need to be built at some future time so there may besome benefits in early site acquisition even if this is not cost effective in the shortterm.

Finally, six-cell sites require refitting of much narrower aperture antennas, but evenwith these in place, parameter setting for cell reselection and handover may be more

difficult.





Multi-Cell Sites

Cost

Visual Impact

Targeted capacity

Advanced features

Evolution to HCS

Long-term solution

Parameter setting

Service

differentiation

More Cells

Figure 19

More Sectors or More Cells?

SC2804/S3/v1.1





4.4 Use of Repeaters

4.4.1 Introduction

Repeaters are bidirectional amplifiers designed to be used in locations wherecoverage from a cell is poor and requires enhancement. They are commonly used toextend more reliable coverage indoors from a donor cell or to fill in coverage holesthat exist because of terrain. In the early rollout phase of a UMTS network they maybe used to increase the general coverage rapidly and at low cost in rural areas,residential areas or along roads and railways.

4.4.2 Donor Antenna Alignment

Repeaters can be made channel selective, but Node Bs in the same HCS layer areseparated only by code. This means that great care must be taken when consideringthe position, type and alignment of the donor antenna for a repeater. Any other NodeB signals arriving at the repeater will also be amplified and reradiated in the repeater area. The result of this could be that UEs in the repeater area are in continuous softhandover with consequential loss of capacity in the system.

It is recommended that the donor antenna should be positioned such that there is at

least an 8 to 10 dB margin between the donor cell signal and other cells in therepeater area. This must then be reflected in handover thresholds.





Node B

Node B

Intendeddonor

Node B

Repeater UE in

repeater area

Risk of continuoussoft handover

Plan for 8 to 10 dBmargin

Figure 20

Repeaters and Antenna Alignment

SC2804/S3/v1.1







ServingNode B

Donor Node B

Disproportionately large powerweighting to compensate for building

loss reduces capacity

Reduced power weighting due torepeater gain increases capacity

Figure 21

Repeaters to Add Capacity

SC2804/S3/v1.1





4.4.4 Antenna Isolation and Gain Setting

The coupling between the two antennas connected to the repeater is referred to as‘isolation in decibels’. Normally the donor antenna will be highly directional. Theantennas in the repeater area may be either omnidirectional or directional.

Any signal coupled from the output of the repeater back to the input via the antennaswill be reamplified. There is a danger that this could lead to positive feedback. Theresulting transmitted noise would have severe implications for capacity in the systemas a whole. It is critical therefore to ensure that the repeater gain is kept below thelevel of isolation to prevent self-oscillation.

It is recommended that in the downlink direction repeater gain is kept at least 15 dBbelow the level of isolation. This margin may be reduced by up to 5 dB in the uplinkdirection. Typically a repeater gain can be set independently in the uplink anddownlink directions up to a maximum of about 90 dB.

Isolation between the antennas should be determined once they are fixed in their final locations. Driving the repeater antenna from a suitably calibrated test transmitter and measuring the power level received at the donor antenna can achieve this. For in-building solutions, the physical structure of the building is interposed between theantennas. This should lead to a better degree of isolation between the antennas than

for outdoor applications.





Repeater

Gain up to c. 90 dB

Donor antenna

Isolation (dB)

Repeater antennaRecommended maximum gain 15 dB less than isolation

Figure 22

Antenna Isolation and Gain Setting

SC2804/S3/v1.1





4.4.5 Node B Desensitization

Consider the repeater shown in Figure 23. Its uplink gain is set to 80 dB and it has anoise figure of 6 dB. Assuming thermal noise of –174 dBm/Hz the noise at the inputto a channel selective repeater when allowing a channel bandwidth of 4.8 MHz willbe –107.2 dBm. This background input level is amplified by 80 dBm and the noisefigure must also be added.

–107.2 + 80 + 6 = –21.2 dBm

Assume an input sensitivity level at the input to the donor cell’s receiver of –102dBm. If the total coupling loss between the repeater output and the donor cell’s inputis less than 80.8 dBm the amplified noise being transmitted back to the donor cell willbe above the threshold of –102 dBm. In these circumstances the repeater isincreasing the noise rise and therefore reducing the cell’s capacity.

The coupling loss includes all antenna gains, the path loss and other forms of gain or loss between the donor cell input and the repeater output. A coupling loss in theorder of 80.8 dB could occur with a spacing between repeater and donor cell of about 400 m, although exact figures will depend on the antennas used and thepropagation path.

If a donor cell and repeater are closely located, then it is worth calculating thecoupling loss and checking in relation to the gain of the repeater whether desensitization seems likely. A reduction of repeater gain may be necessary tocorrect the problem.





Repeater

Donor Node B

–107.2 dBm

80 dBgain

6 dB NF –21.2 dBm

–102 dBm

Gain 80 dB

NF = 6 dB

Coupling loss

Coupling loss = 80.8 dB

Figure 23

Node B Desensitization

SC2804/S3/v1.1





4.4.6 Time Delay in Repeaters

Another consequence of using a repeater is the propagation delay through therepeater. The specific value of delay caused depends on the particular device in use;the value will be available from the vendor as part of the device specification sheet.Typical delays for a repeater will be in the range 5 to 8 µs. A typically delay of 6 µstranslates to a distance travelled of about 1.8 km for a normally propagating radiosignal.

This is not a problem for normal UMTS operation, but it could cause difficulties whentrying to estimate range or position for the UE. For example this would mean that around trip time measurement used to estimate range from a Node B would show anerror of approximately +1.8 km. Also, if using the observed time difference of arrivalmethod for position determination, the error would be at least 900 m and could beconsiderably higher depending on the relative positions of the Node Bs and the UE.

It may also cause a problem if the receiver can see both direct and repeatedversions of the transmitted signal. The delay in the repeater could mean a delayspread greater than the search window for the rake receiver. Thus some channelpaths would be treated as interference.





UE inrepeater

area

Repeater Node B

e.g. 6 µs delayeach way

Round trip time measurement indicates UE is 1.8 km further away than it really is.

May cause problems because of limited search window size in Node B and UE.

Figure 24

Time Delay in Repeaters

SC2804/S3/v1.1





4.5 Basic Considerations for Indoor Coverage

The provision of reliable coverage and sufficient capacity for the indoor environmentpresents specific challenges for the UMTS optimizer. Building loss has a significantimpact on link budgets such that more power must be used by indoor UEs and morepower must be allocated from the Node B. this reduces system capacity.Furthermore, in buildings where there is coverage from more than one outdoor NodeB, UEs are more likely to be engaged in soft handover, thus further reducing systemcapacity.

In addition to this, the indoor environment is much more likely to generate higher data rate traffic. This puts further power and capacity demands on UEs and outdoor Node Bs.

Use of outdoor to indoor repeaters is a cost effective solution that for UMTS canimprove both coverage and capacity. However, capacity gains with this type of solution are limited. The potentially very large traffic density for some in-buildingscenarios and likely different traffic profiles mean that dedicated indoor Node Bs willin many cases be a longer term solution.

There are several options for providing coverage from a repeater or an indoor cell.The primary aim will be to provide sufficient and even coverage across the whole

building area. This may be achieved with a distributed antenna system or with aradiating cable system.





Drivers for Indoor Coverage

indoor UEs need more power and reduce capacity

buildings represent very high traffic densitiespoor coverage in buildings may result in more soft handovers

in-building areas may generate a different traffic profile

expected in mature networks

could use differentiated tariffs

could be a way to compete in different markets

Indoor Coverage Options

repeater

dedicated indoor Node B

copper distributed antenna system

fibre distributed antenna system

radiating cable system

Figure 25

Basic Considerations for Indoor Coverage

SC2804/S3/v1.1





The urban area shown in the diagram measures approximately 2 km in each

direction. The rollout plan assumed a load factor of 65% on each of the cells shown.Each cell has three sets of directional antennas with a maximum gain of 16 dBi.

1 Complete the link budgets using the values shown in this and the next diagramand verify (approximately) that the planned coverage was valid for class 4 UEsrequiring up to 384 kbit/s in the downlink and 128 kbit/s in the uplink. Note thatthe maximum allowed channel code power on the cells is set to -20 dB relativeto the maximum power.

You can convert path loss (Lp) to range in kilometres (d) using COST231-Hata withthe following relationship:

d = antilog Lp – 144.95

37.2

2 Are the cells likely to be uplink or downlink limited?

Customers are starting to complain that they are not able to establish calls in thisarea. The call types and locations vary but the times of day coincide with busiesttimes for these cells.

3 Consider what you think might be happening.

4 Suggest some information you may seek from network statistics to verify your suspicion.

5 Suggest two things you might consider as a solution.

5 EXERCISE 1 – URBAN CAPACITY AND COVERAGE





Antenna gain 16 dBiDiversity gain 3 dB

Duplex filter (1dB)

45 W(46.5 dBm)

Feederloss 3 dB

Antenna gain 0 dBi

UEClass 4

0.125 W (21 dBm)TX

RX

approximately 2 km

Figure 26

Exercise 1 – Urban Capacity

SC2804/S3/v1.1









Uplink

Comments and Conclusions

UE (TX)

Max (TX) power

Antenna gain

EIRP

Node B (RX)

Receiver noise power

in channel

–102.2 dB

Interference margin

Processing gain

Required Eb/No 2.5 dB

Antenna gain

Feeder/connectorlosses

Duplexer loss

Fade margins 10 dB

Diversity gain

Minimum requiredsignal level at antenna

Maximum acceptablepath loss

Downlink

Node B (TX)

Max (TX) power

Feeder/connectorlosses

Duplexer loss

Antenna gain

EIRP

UE (RX)

Receiver noise powerin channel

–102.2 dB

Interference margin

Processing gain

Required Eb/No 1.5 dB

Antenna gain

Fade margins 10 dB

Minimum requiredsignal level at antenna

Maximum acceptablepath loss

Figure 27

Workspace and Proposed Solutions

SC2804/S3/v1.1





6.1 Introduction

The ability to provide location information for a UE is an important aspect of theUMTS system. Location-based services are a potentially significant revenuegenerator for an operator, but additionally the ability to record information regarding aUE’s position may be very helpful for the optimizer.

Location Services (LCS) and information used by optimizers will only be effective if the type and accuracy of positioning information is appropriate to the functionintended. It may be part of an optimizer’s role to consider the different positioningtechniques used and the accuracy of the positioning information.

6.1.1 LCS Clients

There are four categories of LCS Client. These are:

• Value Added Services LCS Clients

• PLMN Operator LCS Clients

• Emergency Services LCS Clients

• Lawful Intercept LCS Clients

Value Added Services LCS Clients use LCS to support VAS, while a PLMN operator may use it to improve operations and maintenance functions or supplementaryservices. By employing LCS the emergency services can assist subscribers whohave made emergency calls. Lawful Intercept LCS Clients may perform services thatare required or sanctioned by law.

6 LOCATION SERVICES (LCS)





Value Added ServicesClients

PLMN Clients

Lawful InterceptClients

Emergency ServicesClients

Police

FBI

E.911

999, 112, 911

weather warnings

enhance network operations

location assisted handover traffic engineering

list of restaurants

places of interestnavigation application

Figure 28

LCS Clients

SC2804/S3/v1.1





6.2 Quality of Service

One aspect of QoS for the geographic location of a UE is accuracy: horizontalaccuracy and vertical accuracy. If appropriate, speed and direction of travel can alsobe taken into account. It may be possible to provide both velocity and geographiclocation, or they can be supplied individually.

Important QoS issues are response time and accuracy. Accuracy of horizontal andvertical data may be considered independently, even where both are requested by aparticular location service.

6.2.1 Horizontal Accuracy

Not all services require the same level of accuracy. For example, the provision of weather reports or traffic information does not need to be pinpointed within a tightgeographical area; it could be an area covering several kilometres. However,tracking information, such as tracking of delivery vehicles or personnel, may needtighter coordinates. Subscribers requiring very localized information, perhaps in atown or city, may need location information that has been calculated down to a fewmetres or tens of metres. The emergency services locating an incident require themost precise information that can be provided.

A range of values is presented to estimate a UE’s position, even for a stationary UE.This is to ensure that the information provided is the best possible within a requiredresponse time. Figure 29 illustrates a range of location services and their estimatedaccuracy requirements.

6.2.2 Vertical Accuracy

It may be possible to provide vertical location information in terms of the actualheight/depth of the target UE, or the estimated height/depth relative to its position at

ground level. Vertical accuracy may range from approximately ten metres tohundreds of metres.





ServiceType

LocationIndependent

Stock prices; sports reports

Services for a particular country or PLMN

Weather reports

Local news; traffic information

People/animal tracking; emergency services;manpower planning

Emergency calls; network-based positioning

SOS; local adverts; ‘where is my nearest?’

Emergency calls; asset location

Emergency calls;route guidance

PLMN/Country

Regional(200 km)

District(up to 1 km)

500 mto 1 km

100 macc. 67%

10–50 m75–125 m 50 m

Figure 29

Horizontal Accuracy

SC2804/S3/v1.1





6.3 Factors Affecting Accuracy of Location Information

While information needs to be as accurate as possible within QoS requirements,several factors may impact upon the provision of a location service. Some of thesefactors are shown in Figure 30 and are described below. The optimization processcan influence most of these factors.

• positioning technique

• geography

• signal attenuation

• multipath propagation and repeaters

• network coverage patterns

There are three defined positioning techniques for UMTS, cell ID based, ObservedTime Difference of Arrival (OTDOA) and network-assisted GPS. The type used maydepend on the QoS requirements and on UE capability.

Geography may affect LCS in a number of different ways. For example, the number and relative position of the base stations; the number of visible satellites, and heightvariation of mobiles and base stations.

When a signal is weakened due to attenuation, it becomes more difficult to makereliable measurements. This is applicable to all the positioning techniques listedabove.

Multipath propagation alters the path length of the signal relative to the geometriclength, giving the impression that the mobile is further away from the base stationthan in fact it is. Perhaps even more significant will be the distortion in measuredpropagation time caused by repeaters. These things are most applicable to the cellID and OTDOA positioning techniques.

The size of a cell may affect LCS, depending on the type of positioning mechanismin use. If a mobile is out of coverage, no positioning information will be available for it.





Factors Affecting Accuracy

Networkcoveragepatterns

Geography

Multipathpropagation

and repeaters

Signalattenuation

UMTS Positioning Techniques:

cell ID basedObserved Time Difference of Arrival (OTDOA)network-assisted GPS

Figure 30

Positioning Techniques and Accuracy

SC2804/S3/v1.1





6.4 Response Time

Response time is an important QoS issue. Different clients and location services willrequire different response times; the issue is particularly important, for example,where a rapid response time is required, as in the case of an urgent positioningrequest.

The response time may be a negotiable QoS parameter. There are three QoSparameters in respect of response times:

• no delay

• low delay

• delay tolerant

When a ‘no delay’ response time is specified, the response will be the Last KnownLocation or Initial Location (if the system holds one for the Target UE). If not, thesystem will return a failure indication and, optionally, initiate procedures to acquire anestimate should it be required. A ‘low delay’ response time will place speed of response above accuracy of information, although it is still with the aim of providingthe greatest possible degree of accuracy relative to the accuracy requirement.However, any attempt at accuracy should not incur additional delay. A ‘delay tolerant’

response time places accuracy above speed. If necessary, a response will bedelayed while the accuracy requirement is fulfilled.

A timestamp will always be provided in respect of location estimates, detailing thetime at which the estimate was obtained.

The network may allocate priority levels to different location services. Requests thatcarry a higher priority level will be processed more quickly than lower-priority ones,and with a greater degree of accuracy. Requests from the emergency services willtake highest priority.





LCS Client

LCS Client

LCS Client

a) No Delay

b) Low Delay

c) Delay Tolerant

LCS

Server

Figure 31

Response Times for LCS

SC2804/S3/v1.1





6.5 Cell ID Based Positioning Mechanism

The positioning mechanism used for a particular request is dependent onUE/UTRAN capabilities and on the QoS requirement for the request. For many typesof service the accuracy provided by simply identifying the cell in which the UE iscurrently located will be sufficient.

In addition to the basic identification of the cell, this position mechanism also allowsfor other information to be included in the position calculation. For FDD mode theRound Trip Time (RTT) may be requested; in TDD mode the receive timing deviationmay be requested. In both cases these can be measured in terms of chip periods.This would enable a range estimation to be made with step increments of about 40m.

6.5.1 Calculation of Geographical Coordinates

If the only information provided for position estimation is the cell ID, then theaccuracy, which in turn will depend on cell size, will depend on the type and locationof the cell. In rural areas cells are likely to have radii greater than several kilometres,but in urban areas cell radii could be measured in hundreds or even tens of metres.

For large and small cells alike the UE’s position is only identified within the coveragearea of the cell. Where geographical coordinates are required as the response to thelocation request, a default position within the cell must be defined. It would makesense for this to be the geographical centre of the cell. However, knowledge of trafficdistribution within the cell (for example if it was covering a major road or included abusy shopping street) could be taken into account when defining a default location. If distance information is included, then defined default position may take the form of aline across the cell.

6.5.2 UE State

At the time of the location request the UE’s associated cell ID may or may notalready be known, depending on the current RRC state of the UE. If the UE has acurrent RRC connection a cell ID may already be known for the UE. However, if theUE is in the URA_PCH state or if it has no RRC connection and is in idle mode, acell ID will not be known. For UEs in the URA_PCH state a transition to theCELL_FACH state can be forced by paging initiated by the SRNC. If a UE is in idlemode, paging will need to be initiated from the core network.





UE

X

Default location

for coordinates

Error margin

Node B

RTTmeasurements

Figure 32

Cell ID Based Positioning

SC2804/S3/v1.1





6.5.3 Cell ID During Soft Handover

When the UE is in soft handover it will be associated with more than one cell, so astrategy is required to determine which cell ID should be used for indicating the UE’sposition. Several methods are suggested, including:

• quality measurements of cells in the soft handover

• the cell on which the call was set up

• the cell provided by the closest Node B to the UE

• the cell most recently added to the soft handover

In the rollout phase it is likely that one policy will be adopted for all location requests.However, as the network matures the policy could be refined through optimization.

In the example shown in Figure 33 the UE is in a three-way soft handover. It isclosest to Node B 1, but the call was first set up on Node B 3 and Node B 2 wasmost recently added to the soft handover.

Which of these may be most appropriate to use as a cell ID for location coulddepend on why the location is being requested. Node B 1 may best represent theUE’s physical location, but Node B 2 may give a better indication of where the UE isgoing. However, the call was set-up on Node B 3, and the location request may beassociated with a service related to call establishment.





Closest

Call Set-upNewest Added

– Node B 1

– Node B 3 – Node B 2

UE

Node B1

Node B2

Node B3

Figure 33

Cell ID During Soft Handover

SC2804/S3/v1.1





6.6 Observed Time Difference Of Arrival (OTDOA)

6.6.1 Hyperbolic Position Calculation

This mechanism involves the UE taking measurements of the OTDOA between thedownlink transmissions from pairs of Node Bs. As shown in Figure 34, a constantmeasured time difference between the downlink signals from Node B 1 and Node B2 describes a line. The line takes the form of a hyperbolic curve and is known as aLine Of Position (LOP).

In general, to estimate the position of a UE in a two-dimensional plane,measurements are required from at least two pairs of Node Bs, i.e. a minimum of three Node Bs are involved. This provides two intersection curves; this is known astrilateration. However, it is possible for two curves to have two points of intersection.In such cases it is necessary to add measurements from a third pair of Node Bs inorder to give an unambiguous position.

A software function called the Position Calculation Function (PCF) translatesmeasurements into position coordinates. In UE-assisted mode the PCF is resident inthe Serving RNC (SRNC), with the UE only returning measurement results. In UE-based mode the PCF is resident in the UE. The system sends assistance data to theUE, which then performs both the measurements and the calculation in order to

return the position coordinates to the SRNC.

It is possible to use the OTDOA mechanism to derive a three-dimensional positionfor the UE. To do this it is necessary to consider a plane of constant difference rather than a line of constant difference. This plane will be hyperboloid in shape. Twohyperboloid planes will not provide an unambiguous position since their intersectionwill be elliptical. If three hyperboloids are identified then their two ellipticalintersections may provide a unique point. Ideally, however, four hyperboloid planeswould be used.





Node B2

Node B1

Node B3

Example UE positiondescribed by two

Lines Of Position (LOP)

LOPs representing equal distance:

Node B 1 to Node B 2

Node B 1 to Node B 3

Figure 34

OTDOA Mechanism

SC2804/S3/v1.1





6.6.2 Position Accuracy for OTDOA

There are many factors that will influence the accuracy of this positioningmechanism, including:

• measurement resolution

• measurement accuracy in the UE

• radio channel and propagation conditions

• accuracy of the Node B’s known position

• relative position of the Node Bs

The measurement resolution for time difference is the chip period of 260 ns. Thisequates to approximately 40 m difference between the LOP curves for a given pair of Node Bs. As shown in Figure 35, noise, propagation conditions, measurement errorsand measurement quantization will result in a Probability Distribution Function (PDF)surrounding the assumed exact position of the LOP. If the bounds of the PDF aretaken to be σ then it represents 68% confidence.

In Figure 35 this has been done for LOP1 and LOP2. At their intersection the overlapof the two PDFs and the probability of the UE’s position being contained within it is

the product of the two PDFs, i.e. 46.6%. It is possible to construct an ellipse insidethis area with axes ‘x’ and ‘y’, which gives a good indication of the effects of measurement error.

To maintain 68% confidence it is necessary to construct a circle that has a radiusequal to the square root of x2 + y2. It is then apparent that the size of this circle is afunction of the angle of intersection between the two LOPs. Thus the closer their intersection is to a right angle, the more confidence there can be in the UE’s position.

This is an important factor when deciding which Node Bs are used for measurements. Ideally, it should also be a consideration when selecting Node B

sites, but in practice there are other, more inflexible requirements that drive siteselection. Therefore, to increase the level of accuracy, it is possible to install anetwork node called a Location Measurement Unit (LMU). These produce signals theUEs can use for measurements and can be placed with regard only to improvingmeasurement accuracy.





68% confidence

x

y

PDF for LOP 1

PDF for LOP 2

LOP 1

LOP 2

r

r = x2 + y2

Figure 35

Position Accuracy for OTDOA

SC2804/S3/v1.1





It can also be seen that the space between the adjacent LOP curves increases with

increasing distance from the centre of each curve. This is known as the LaneExpansion Factor and is another reason why the relative positions of the Node Bsand LMUs is important.

The degradation in accuracy due to the relative position of measured transmitters issometimes called ‘geometric dilution of position’.

Radio propagation conditions may have a significant effect on positioning accuracy.Reflections and diffractions will increase the path length relative to the geometricdistance between the UE and the Node B. The degree to which this affects accuracyis a factor of the cell’s location. In order to translate time differences into a position, itis necessary to have accurate positional information for the Node Bs. This meansthat the location of the transmitters must be accurately surveyed. It is worth notingthat the position must be that of the electrical centre of the transmitting antenna andnot the position of the Node B. If an antenna array is being used for beam steeringthen this point may change and, depending on the accuracy required, may need tobe accounted for in the position calculation.





68% confidence

x

y

PDF for LOP 1

PDF for LOP 2

LOP 1

LOP 2

r

r = x2 + y2

Figure 35 (repeated)

Position Accuracy for OTDOA

SC2804/S3/v1.1





6.6.3 Real Time Difference (RTD)

In order to perform the position calculation, account must be taken of the Real TimeDifference (RTD) between downlink transmissions. UMTS TDD systems are usuallyfully synchronized, so RTD will be a constant value, which can be entered into adatabase. However, UMTS FDD systems are non-synchronized.

The non-synchronization between FDD Node Bs means that the RTD between NodeBs will slowly drift. For example, it would be possible for the RTD between two NodeBs operating within specified tolerance to drift by one chip period in about 2.5 hours.

In a non-synchronous system it is the function of the LMUs to measure and updatevalues of RTD between Node Bs. The updated values are then passed to the SRNC.





Node B2

Node B1

UE

t

Ref

t

Ref RTD

t

OTD

RTDGTD

Geometric

Geometric Time Difference The difference between the reception of signals from twostations due to geometry.

–

Observed Time Difference The timing difference between Node B 1 and 2 as measuredat the UE.

–

Real Time Difference Local time at Node B 1 – local time at Node B 2. –

Figure 36

Real Time Difference (RTD)

SC2804/S3/v1.1





6.6.4 Use of Idle Periods

In order to take measurements of observed time difference a UE must be able tohear neighbour cells. In a CDMA-based system this is very difficult for a UE near aNode B because of traffic and signalling transmissions to other UEs in the cell. Thisis known as the ‘hearability problem’. This is solved by a mode of operation knownas Idle Period Downlink (IPDL). In this mode of operation the Node B periodicallyceases downlink transmission on all channels. This provides UEs with a silenceperiod during which they can take reliable measurements of neighbour cell timing.The Node B informs UEs of idle periods in higher-layer signalling.

The UMTS specifications1 contain a number of parameters that are used to definehow idle periods will be operated in a cell. It is possible that these parameters couldbe subject to optimization activity.

1 3GPP TS 25.214 Physical Layer Procedures (FDD).





Strongsignal

Weaksignal

Node B2

Node B1

UE

Idle Idle Serving Cell

Neighbour Cell

t

Figure 37

Use of Idle Periods

SC2804/S3/v1.1





6.7 Network-Assisted Global Positioning System (GPS)

The GPS is a constellation of low earth orbiting satellites operated by the USDepartment of Defence. It is very widely used for providing timing and positioninginformation in varied applications. GPS satellites transmit synchronized CDMAsignals, timing and constellation position information. This enables a terminal tocalculate a three-dimensional position by measuring the observed code phase shiftfor several (ideally at least four) satellites.

The reduction of cost and size of GPS reception equipment makes it feasible for it tobe incorporated into a UE. Additionally the selective availability feature, which limitedcivilian access to an accuracy of 100 m, was turned off in May 2000. This means thatin the right conditions this method can offer a high degree of vertical and horizontalaccuracy. This may be operated in either UE-assisted or UE-based modes. In theUE-based mode the UE contains a full implementation of the GPS receiver so that itcan perform both the measurements and position calculation internally. In the UE-assisted mode the UE can contain a simpler, limited-function GPS receiver so that itis able to carry out timing measurements only. The measurement results are thenreturned to the SRNC, where the position calculation is carried out.

6.7.1 Network Assistance

There are some disadvantages with using GPS. These include unreliability in weak-signal cases (in-building), long acquisition time and very high power consumption,particularly while a fix is being taken. For its application in UMTS, GPS assistancedata is provided for the UE in order to alleviate some of these problems. Thisassistance data includes information about satellite visibility, timing and position. Theaim is to improve performance in terms of position calculation accuracy, reducedacquisition time, lower power consumption and improved performance in low signalstrength conditions.





UE

Assistance Data

Node B

SRNC

GPS

GPS

GPS

Figure 38

Network Assisted GPS

SC2804/S3/v1.1





There are many widely accepted propagation models that have been used in the

planning and optimization of GSM and other 2G systems. However, UMTSintroduces a need for new levels of accuracy in propagation prediction. This isbecause the application of CDMA within UMTS makes it very sensitive to smallchanges. In particular there is a high coupling between channels and cells such thatthe behaviour of one Node B or even one UE can have wide-reaching effects onlarge geographical areas. Additionally, there are the large number of parametersrequired to control system access, power control and handover. The mutualinteraction between channels and cells means that it is important for the optimizer tobe confident that predictions used for simulating optimization changes are anaccurate representation of real system conditions.

There are two main categories of propagation model, empirical models anddeterministic models.

7.1 Empirical Models

Empirical models are based on a power law modified to align with best-fit curvesderived from real-word measurements. Perhaps the best known of these is theOkumura–Hata model. The COST231-Hata model is based on measurements takenin several modern European cities and is a development of the Okumura–Hata

model. It is widely used and generally considered to be suitable for planning UMTSmacro cells. The urban variant of COST231-Hata is shown in Figure 39,modifications for suburban, quasi-open and open areas are also available.

It is very important to tune an empirical model to suit the specific location in which itis to be used. Ideally this should be done for every cell. In practice this would be verycostly and may be impossible during the initial planning and rollout phase for a newUMTS network. Nevertheless, great accuracy is required for effective UMTSoptimization.

7 PROPAGATION MODELLING





xx

x

x

x

x

x

x

x

x

xx

xx

x

x

x

xx

x

x

x

xxx

COST231-Hata

Lp (urban) = 46.3 + 33.9Logf – 13.82Loghb – a(hm) + (44.9 – 6.55Loghb)Logd + Cm

Where

hb and hm are in meters, d is in kilometres and f is in MHz.

and a(hm) = 3.2(Log(11.75hm))2 – 4.97 for a large city

or a(hm) = (1.1Logf – 0.7)hm – (1.56Logf – 0.8) for a small to medium city

and Cm is 3 dB for metropolitan centres and 0 dB for medium sized cities or suburban areas.

frequencyantenna heightregion type

Variables

Signallevel

DistanceBest-fit curve

Figure 39

Empirical Models: COST231-Hata

SC2804/S3/v1.1





7.1.1 Accuracy for Simulations

The potential for making system performance worse with a change that has not beenaccurately simulated is significant in UMTS, and would be far more likely to occur without accurate propagation prediction. Therefore, accurate radio measurementsand tuning of the radio model should ideally be performed on established cells beforeoptimization activity begins. This will mean that more accurate simulations can beperformed to check the likely impact of proposed optimization solutions.

A number of comparisons have been made between the performance of empiricalmodels and that of deterministic models. In general, they work well in open areas butwith degraded performance in urban areas. Even when well tuned, the effects of street canyons and building loss means they can exhibit considerable localizederrors. These errors are tolerable in a TDMA-based system such as GSM, butprobably not in a CDMA-based system such as UMTS. This suggests that moreaccurate modelling methods should be used for UMTS, at least for optimizationpurposes.





xx

x

x

x

x

x

x

x

x

xx

xx

x

x

x

xx

x

x

x

xxx

COST231-Hata

Lp (urban) = 46.3 + 33.9Logf – 13.82Loghb – a(hm) + (44.9 – 6.55Loghb)Logd + Cm

Where

hb and hm are in meters, d is in kilometres and f is in MHz.

and a(hm) = 3.2(Log(11.75hm))2 – 4.97 for a large city

or a(hm) = (1.1Logf – 0.7)hm – (1.56Logf – 0.8) for a small to medium city

and Cm is 3 dB for metropolitan centres and 0 dB for medium sized cities or suburban areas.

frequencyantenna heightregion type

Variables

Signallevel

DistanceBest-fit curve


Empirical Models: COST231-Hata

SC2804/S3/v1.1





7.2 Deterministic Models

These are physical models based on knowledge of wave theory and on detailedknowledge of the morphological and electrical characteristics of the localenvironment. The most widely used deterministic technique is ray tracing. Raytracing models calculate specific reflections and diffractions for rays launched intothe modelled environment. The aim is reproduce as closely as possible real-worldpropagation. The most accurate prediction comes with three-dimensionalenvironmental data, but two-dimensional predictions can also be effective in someenvironments.

For ray tracing to be effective it is necessary to have accurate data about theenvironment to be modelled. This kind of data is now more widely available, whichmakes ray tracing more viable. Another limiting factor for ray tracing in the past hasbeen the lack of sufficient processing power for it to be performed on a large scale.In recent years this too has become a much less significant problem.

A number of trials have shown that ray tracing is significantly more effective for predicting signal level in urban and in indoor areas than empirical techniques. Thismakes it a much more effective tool for the optimizer.





Node B

Figure 40

Deterministic Models: Ray Tracing

SC2804/S3/v1.1





7.3 Comparing Models and Their Effects

It is possible to compare real measurements with predictions based on empirical anddeterministic models in the most effective way to assess their accuracy. Studies inwhich this has been done suggest that the impact of buildings (canyon effects andpenetration loss) are not well modelled by empirical methods. In general, this meansthat empirical models tend to underestimate signal level in streets at the edge of acell’s predicted coverage area. The consequence of this is that they underestimatethe overlap for adjacent cells. Dependent on street layout this error can be extreme.This concept is illustrated in Figure 41.

The likely result of this error is that the real network will show a higher level of interference and greater occurrence of soft handover than the simulations suggest.Thus the real network will have less capacity than the simulated network. It followsthat the pessimistic coverage estimates of empirical models may result in a cell plancontaining more cells than necessary. This raises the probability that the conclusionof an optimization study may be to suggest the removal of cells in order to reduceinterference and thus increase capacity. The need to do this would only becomeevident with very accurate coverage predictions. Indeed, simulations performedusing inaccurate coverage predictions could lead to optimization changes thatdegrade system performance rather than improve it.





Small predictedoverlap area

Typical empirical model results

Large predictedoverlap area

p deterministic model results

Figure 41

Comparing Models

SC2804/S3/v1.1









SECTION 4

RAN CONFIGURATIONS AND

DIMENSIONING

i









1 UMTS Channels 4.11.1 Access Stratum (AS) 4.11.2 Non-Access Stratum (NAS) 4.11.3 The AS on the Air Interface 4.31.4 Logical Channels 4.51.5 Transport Channels 4.7

1.6 Downlink Physical Channels 4.111.7 Uplink Physical Channels 4.151.8 Channel Mapping Options 4.17

2 Cell Configuration 4.192.1 Example Downlink Channels 4.192.2 Example Uplink Channels 4.192.3 Configuration Options 4.212.4 Using More than One Frequency 4.23

3 Cell Transmit Power 4.25

3.1 Downlink Power 4.253.2 Downlink Power Weightings 4.273.3 Varying the CPICH Weighting 4.293.4 Utilizing Soft Capacity and Dynamic CPICH Power 4.313.5 Pilot Pollution 4.33

4 Antenna Configurations 4.354.1 Use of Downtilt 4.354.2 Calculations for Beamtilt 4.374.3 Practical Antenna Types and Tilt Effects 4.394.4 Choice of Antenna 4.53

5 Radio Performance 4.555.1 Minimum Coupling Loss 4.555.2 Adjacent Channel Leakage Ratio (ACLR) 4.575.3 Radio Carrier Spacing 4.595.4 Adjacent Channel Interference (ACI) 4.615.5 Reducing ACI 4.63

6 Interaction and Interference with GSM 4.656.1 Transmitter Noise and Spurious Emissions 4.65

6.2 Receiver Blocking 4.69

SECTION CONTENTS

iii










• list the logical, transport and physical channels applicable to UMTS FDD

mode

• describe FDD mode channel mapping options, channel characteristics and

traffic applicability

• describe typical UMTS cell configurations for a range of traffic scenarios

• describe and justify options for downlink power weightings

• identify how different antenna configurations can be used to optimize

coverage and capacity

• identify appropriate antenna types and configurations for a range of cell

types

• describe how an operator may use multiple carrier allocations to optimize


• describe downlink channel power allocations and limitations in terms of


• describe the impact of Adjacent Channel Leakage Ratio (ACLR) and

describe optimization options to combat it

• identify how capacity and coverage may be limited by spurious emission and

receiver blocking characteristics

SECTION OBJECTIVES

v





Isolation of radio-related functions from the data networking functions is achieved by

splitting the air interface into two distinct areas: the Access Stratum (AS) and theNon-Access Stratum (NAS).

1.1 Access Stratum (AS)

The AS provides communication between the UE and the UTRAN, managing theUMTS radio interface and providing services, called Radio Access Bearers (RAB), tothe NAS. The AS can be considered as being layers 1–2 of the OSI Seven-Layer Model, with some layer 3 functionality. The main AS functions are:

• provision of physical channels

• control of physical channels

• link establishment and clearing

• channel coding

• some security functions

1.2 Non-Access Stratum (NAS)

The NAS provides communication between the UE and the Core Network (CN). TheNAS acts transparently through the UTRAN and can be considered as being carriedby, rather than being, the air interface. The NAS can be considered as providinglayers 3–7 of the OSI Seven-Layer Model. The NAS is used to invoke and provideoverall control of a number of air interface procedures.

1 UMTS CHANNELS





L7

L3

Access

Stratum

Non-Access Stratum

UTRAN

Core NetworkOSI Layers

L3

L1

Uu

Relay

UE

Iu

Figure 1

UTRAN Architecture

SC2804/S4/v1.1





1.3 The AS on the Air Interface

The AS covers functionality from layers 1–3. At layer 1, signalling and traffic data iscarried across the air interface in physical channels that are defined in terms of either code set and frequency for FDD mode, or code, timeslot and frequency for TDD mode.

Layer 2 is divided into two sublayers. The lower sublayer is the Medium AccessControl (MAC) layer. It is responsible for a wide range of functions including randomaccess procedures, physical link control, multiplexing and channel mapping to thephysical layer. The upper sublayer is the Radio Link Control (RLC) layer, which isresponsible for Logical Link Control (LLC) and acknowledged and unacknowledgeddata transfer. Ciphering may be provided by either RLC or MAC. Layer 3 in the ASprovides only the lower part of layer 3 in the control plane. This is known as theRadio Resource Control (RRC) layer. It is responsible for the coordination andcontrol of a range of functions including bearer control, monitoring processes, power control processes, measurement reporting, paging and broadcast control functions.

1.3.1 Logical and Transport Channels

There is a complex array of user and signalling requirements. In order to define a

process for each type of information, sets of logical channels mapping into transportchannels and ultimately to physical channels are defined. Logical channels aredefined between RLC and MAC. Transport channels are defined between MAC andthe physical layer.





Control Plane Signalling

Radio ResourceControl (RRC)

Radio LinkControl (RLC)

Medium AccessControl (MAC)

Physical Layer

Transportchannels

Logical

channels

L2

L3

L1

User Plane Information

Figure 2

AS on the Air Interface

SC2804/S4/v1.1





1.4 Logical Channels

The MAC layer provides transfer services via a set of logical channels. A logicalchannel is defined for each different transfer requirement. Each logical channelrelates to particular kinds of information that need to be transferred. Some relate tosignalling information, and some to traffic information.

The logical channels used for the transfer of signalling information in FDD mode arethe Broadcast Control Channel (BCCH), Paging Control Channel (PCCH), CommonControl Channel (CCCH) and Dedicated Control Channel (DCCH).

The logical channels used for the transfer of user information in FDD mode are theDedicated Traffic Channel (DTCH) and the Common Traffic Channel (CTCH).

1.4.1 Logical Channel Types

Broadcast Control Channel (BCCH)The BCCH is a downlink broadcast channel carrying system information.

Paging Control Channel (PCCH)The PCCH is a downlink channel carrying paging messages. It is used when the

network does not know the location cell of the UE, or the UE is using sleep modeprocedures.

Common Control Channel (CCCH)This is a bidirectional channel carrying control information between the network andthe UE. It is used when the UE has no RRC connection with the network.

Dedicated Control Channel (DCCH)This is a point-to-point bidirectional channel carrying dedicated control informationbetween the network and the UE. It is used when a dedicated connection has beenestablished through RRC connection set-up procedures.

Dedicated Traffic Channel (DTCH)The DTCH is a dedicated point-to-point channel carrying user information betweenthe network and the UE. It may be used in both the uplink and downlink directions.

Common Traffic Channel (CTCH)The CTCH is a common point-to-multipoint downlink-only channel used for carryingbroadcast or multicast user information.





Medium Access Control (MAC)

Control Channels from RLC

Traffic Channels from RLC

BCCH PCCH CCCH DCCH

DTCH CTCH

Figure 3

Logical Channel Types

SC2804/S4/v1.1





1.5 Transport Channels

Information is transferred from the MAC layer and mapped into the physical channelsvia a set of transport channels. Transport channels can be classified into two groups:common channels and dedicated channels. Information in common channels willrequire in-band identification of the UE. For dedicated channels the UE’s identity isassociated with the channel allocation.

The common transport channels for FDD mode are the Random Access Channel(RACH), Common Packet Channel (CPCH), Forward Access Channel (FACH),Downlink Shared Channel (DSCH), Broadcast Channel (BCH) and the PagingChannel (PCH).

The dedicated transport channel for FDD mode is the Dedicated Channel (DCH).

1.5.1 Transport Formats

Each transport channel has an associated transport format. This is defined as acombination of encoding, interleaving, bit rate and mapping into physical channels.For some transport channels this may be variable within a set of transport formats.





DCH

Physical Layer

Common Channels from MAC

Dedicated Channelsfrom MAC

RACH CPCH FACH DSCH BCH PCH

Figure 4

Transport Channel Type

SC2804/S4/v1.1





1.5.2 Transport Channel Types

Random Access Channel (RACH) A contention-based channel in the uplink direction, the RACH is used for initialaccess or non-real-time dedicated control or traffic data.

Common Packet Channel (CPCH)This channel is only used in FDD mode. It is a contention-based channel used for the transmission of bursty traffic data in a shared mode. Fast power control is used.

Forward Access Channel (FACH)The FACH is a common downlink channel without power control. It is used for controlor traffic data.

Downlink Shared Channel (DSCH) A downlink channel used in shared mode by several UEs, the DSCH is used to carrycontrol or traffic data.

Broadcast Channel (BCH)This is a downlink broadcast channel used to carry system information across awhole cell.

Paging Channel (PCH)The PCH is a downlink broadcast channel used to carry paging and notificationmessages across a whole cell.

Dedicated Channel (DCH)The DCH is used in the uplink or downlink direction to carry user information to or from the UE.





DCH

Physical Layer

Common Channels from MAC

Dedicated Channelsfrom MAC

RACH CPCH FACH DSCH BCH PCH

Figure 4 (repeated)

Transport Channel Type

SC2804/S4/v1.1





1.6 Downlink Physical Channels

In the downlink direction there are a number of channels carrying higher-layer information and a large number having control and synchronization functionsassociated with layer 1.

1.6.1 Physical Downlink Shared Channel (PDSCH)

This is a DL channel used to carry the DSCH. It is shared by multiple users by way of code multiplexing. The PDSCH is always associated with one or more DL DedicatedPhysical Channels (DPCHs).

1.6.2 Secondary Common Control Physical Channel (SCCPCH)

The SCCPCH is used to carry the transport channels PCH and FACH in the DLdirection. There may be one or more SCCPCHs, and if an SCCPCH is only carryingthe FACH, it may be transmitted over only part of the cell using beam-formingantennas.

1.6.3 Primary Common Control Physical Channel (PCCPCH)

This is used in the downlink direction to broadcast the BCH across a cell. There willbe only one of these on each cell.

1.6.4 Dedicated Physical Data Channel (DPDCH) and Dedicated PhysicalControl Channel (DPCCH)

The DPDCH is a bidirectional channel used to carry higher-layer information from thetransport channel DCH. It is multiplexed with the DPCCH that provides the layer 1

control and synchronization information. Once multiplexed, the two are referred to asa DPCH. One DPCCH may be associated with one or more DPDCHs

1.6.5 Paging Indicator Channel (PICH)

This DL channel is used to carry Paging Indicators (PI). These are used to enablediscontinuous reception of the PCH being carried on an associated SCCPCH.





PICHSCHCPICH AICH AP-AICHCD/CA-ICHCSICH DPCCH

DPCH

DPDCH

DCH

PCCPCH

BCH

SCCPCH

FACH PCH

PDSCH

DSCH

Transport Channels

Layer 2

Layer 1

Physical Channels

Figure 5

Downlink Physical Channels

SC2804/S4/v1.1





1.6.6 Synchronization Channel (SCH)

This is a downlink channel used during cell search. It consists of primary andsecondary subchannels, and conveys information to the UE concerning the timealignment of a cell’s codes and frame structures.

1.6.7 Common Pilot Channel (CPICH)

This channel is used to provide the phase reference for the SCH, PCCPCH, AICHand the PICH. It may also be the default phase reference for all the other DLchannels. There will be only one Primary CPICH in a cell. It is an option to have oneor more Secondary CPICHs in a cell. If present, the Secondary CPICHs would act asthe phase reference for SCCPCHs, and potentially DPCHs.

1.6.8 Acquisition Indicator Channel (AICH)

This downlink channel carries Acquisition Indicators (AI). These are used toacknowledge UE random access attempts, and grant permission for a UE tocontinue with its random access transmission.

1.6.9 Physical Channels for Common Packet Channel (CPCH) Access

These channels carry information used for the CPCH access procedure and do notcarry transport channels.

CPCH – Access Preamble Acquisition Indicator Channel (AP-AICH)This channel carries AP acquisition indicators that correspond with the AP signaturetransmitted by the UE. It is also used to acknowledge the random access preambles,which are then followed by a collision detection preamble.

CPCH – Collision Detection/Channel Assignment Indicator Channel(CD/CA-ICH)The CD/CA-ICH is used to acknowledge the collision detection access preamble.

CPCH – Status Indicator Channel (CSICH)The CSICH uses the unused part of the AICH channel to indicate CPCH physicalchannel availability so that access is only attempted on a free channel.





PICHSCHCPICH AICH AP-AICHCD/CA-ICHCSICH DPCCH

DPCH

DPDCH

DCH

PCCPCH

BCH

SCCPCH

FACH PCH

PDSCH

DSCH

Transport Channels

Layer 2

Layer 1

Physical Channels

Figure 5 (repeated)

Downlink Physical Channels

SC2804/S4/v1.1





1.7 Uplink Physical Channels

In the UL direction there are three types of physical channel: the Physical Random Access Channel (PRACH), Dedicated Physical Channel (DPCH) and the PhysicalCommon Packet Channel (PCPCH)

1.7.1 Physical Random Access Channel (PRACH)

This UL channel is a contention-based channel used to carry higher-layer information in the form of the RACH.

1.7.2 Dedicated Physical Channel (DPCH)

The DPCH is ultimately used to carry the transport channel DCH. However, inaddition to this it carries layer 1 information in the form of the pilot, Transmit Power Control (TPC), and Transport Format Combination Indication (TFCI) bits.The DPCHcan therefore be considered as two subchannels: the DPDCH, which is used to carryDCH; and the DPCCH, which is used to carry the layer 1 information. These twosubchannels are multiplexed together to form the DPCH.

1.7.3 Physical Common Packet Channel (PCPCH)

The PCPCH carries the common packet transport channel, which comprises accesspreambles, collision detection preamble, power control preamble and a messagepart.









BCCH PCCH DCCH CCCH CTCH DTCH

BCH PCH CPCH RACH FACH DCHDSCH

PCCPCH SCCPCH PCPCH PRACH PDSCH DPCH

Air

Interface

Physical

Layer

Transport

Channels

Logical

Channels

MAC

Figure 7

Channel Mapping Options

SC2804/S4/v1.1





Figure 8 shows how a typical UMTS cell may be configured in the uplink and

downlink directions.

2.1 Example Downlink Channels

The cell will contain a single PCCPCH. This channel carries the BCH transportchannel, which in turn carries system information messages. Phase synchronizationfor this physical channel is provided by the CPICH. These channels will always bescrambled by the cell-specific primary scrambling code. The two channels will betime-aligned in terms of scrambling code and frame structure, this timing beingindicated by the Primary and Secondary SCH.

This example cell contains one SCCPCH. This channel is being used to carry theFACH and PCH. These are variable-rate channels that, in the case of FACH, maycontain a mixture of signalling and traffic.

There are several types of physical channels with which a cell may be provisionedthat carry only physical layer signalling. Two of these are shown in Figure 8: the AICH, which is used to acknowledge random access probes, and the PICH, which isused to support a discontinuous reception function for the PCH.

There are likely to be multiple DPCHs and PDSCHs in operation on the cell. Theseare variable-rate channels that may carry signalling or traffic. In general, burstypacket-switched traffic is likely to be carried in the DSCH, while circuit-switchedtraffic must be carried in a DCH.

2.2 Example Uplink Channels

In the UL direction there are three physical channel types with slightly different coderequirements. The PRACH and the PCPCH are always directed at a single cell; softhandover is not a feature of these channels. As a result, the codes used can be cell-

specific. Up to 16 PRACH and up to 64 CPCH channels could be provisioned on acell but the example cell has two PRACHs and four PCPCHs.

The cell also has provision for uplink DPCHs to match those operating in thedownlink direction. These channels can use soft handover and therefore the codesare not cell-specific.

2 CELL CONFIGURATION





PRACH(multiple e.g. 2)

PCPCH(multiple e.g. 4)

DPCH(multiple)

RACH(multiple e.g. 2)

CPCH(multiple e.g. 4)

DCH(multiple)

BCH

PCHFACH

SCH

CPICH

DPCH(multiple)

PICH AICH

PCCPCH

PDSCH(multiple)

SCCPCH

DCH(multiple)DSCH

(multiple)

Downlink

Uplink

Figure 8

Example Cell Configuration

SC2804/S4/v1.1





2.3 Configuration Options

There are many cell configuration options available to a UMTS operator. The mostsuitable configuration will depend on location and the likely traffic profile of users inthe cell area. At rollout it is likely that all cells will be configured in one way only or perhaps in a limited set of default ways.

Optimization can include consideration of channel provisioning on a cell. Given thatdifferent channels are suited to different traffic characteristics it is likely that thechannel types available on a cell could be optimally matched to the local trafficrequirements.

For example, a cell being used for an indoor coverage solution is more likely to carryhigh-bit-rate packet-switched data. This means that more CPCHs and DSCHs mayneed to be provisioned. In addition to the changes to the site database this will alsoimpact the Node B’s physical requirements for channel elements and terrestrialtransmission bandwidth.

Another possibility is to build a new cell to provide a specific function. For example,at a sports stadium or in a large public arena a cell could be used to stream audioand visual content, perhaps as a commentary of an event. This would require theCTCH, which is mapped into the FACH. This could be operating at a very high bit

rate requiring the construction of a cell with its capacity predominantly dedicated to aSCCPCH carrying the FACH.





Standardconfiguration

In-building – moreprovision for packetdata with CPCH and

DSCH

Sports stadium –more provision for

streamed audio/videowith CTCH in FACH

Figure 9

Configuration Options

SC2804/S4/v1.1





2.4 Using More than One Frequency

Most UMTS licence holders have enough radio spectrum for more than one FDDradio carrier pair. It is possible to add second or even third frequencies to a Node B.However, this concept is slightly different in UMTS than in GSM.

A Node B can contain one or more cells. A typical arrangement would be to havethree cells using appropriate directional antenna on a Node B site. All three cellswould be using the same frequency. It would be possible to add more capacity to theNode B by adding a second frequency for each set of antennas. However, each newfrequency added carries its own full set of control and traffic channels. This meansthat the second frequency must be considered as a new cell. Thus a three-cell NodeB becomes a six-cell Node B even though only three sets of directional antennas areused.

It is possible to use wideband power amplifiers so that a single power amplifier canamplify two frequencies. This would save cost because although the site has sixcells, only three power amplifiers would be needed. However, this means that thepower available to each cell is halved. If the cells are downlink limited then this willhalve the capacity of the cells.





F2Cell 4

F2Cell 5

F2Cell 6

Three sets of antennasThree cells

One frequency

Three sets of antennasSix cells

Two frequencies

F1Cell 1 F1

Cell 2

F1Cell 3

F1Cell 1 F1

Cell 2

F1Cell 3

Figure 10

Using More than One Frequency

SC2804/S4/v1.1





3.1 Downlink Power

In the downlink direction the maximum transmit power available from the highly linear power amplifier can be considered constant. The total power available will depend onthe vendor and on the type of Node B. For a macro cell product it could be expectedto be in the range 20 to 45 W, for a micro cell or pico cell product it would beproportionally lower. The specifications1 require that the power amplifier has a totalpower dynamic range of at least 18 dB. Maximum transmit power is limited to 50dBm (100 W).

This power is shared between all downlink channels. Downlink power control isimplemented through the adjustment of the weighted sum of the downlink channels.Broadcast and common control channels are likely to be allocated a fixed proportionof the power available. The remainder of power is then shared between users. Theweighting may be used to vary proportions to each user dependent on path loss,interference and required quality of service. For closed loop power control the UEindicates the requested power step changes to the Node B. However, a limit will beset for the power proportion available to each channel type, so the Node B may notobey all power control commands. If the cell is operating at less than full load thenthe total power transmitted is less than the total power available.

More power is required if:

• there are more channels required

• if users are distant from the Node B

• if users request higher data rates

• if users request a higher quality of service

Thus the total power available in a cell ultimately limits downlink capacity and qualityof service.

1 3GPP TS 25.104 BS Radio Transmission and Reception (FDD).

3 CELL TRANSMIT POWER





G1

Modulation andlinear power

amplifier

G2

Gn

Channel 1

Channel 2

Channel n

Nearby user

Nearby user

Distant user

SCCPCH

PCCPCH

CPICH

Maximumtransmitpower

Totaltransmitpower

0

Currentlyunused

Higher weighting

Lower weighting

Dynamic rangeat least 18 dB

Figure 11

Downlink Power

SC2804/S4/v1.1





3.2 Downlink Power Weightings

The power weighting allocated to downlink broadcast and common channels, alongwith the range for power control available to other channels, is an important part of acell’s configuration.

All downlink channels will be allocated a fraction of the maximum transmit power available on the cell. This fraction is referred to as the channel code power and canbe in the range –3 dB to –28 dB relative to the maximum transmit power available.Thus the maximum proportion that can be allocated to any individual channel is 50%of the maximum power.

The starting point and reference for all other channels in the CPICH. The standardsallow for the code power in this channel to be set from –10 dBm to 50 dBm.However, the important consideration for this channel is the percentage allocated toit from the total power available for the cell. A typical value for this percentage is 10%of total transmit power. Nevertheless, the optimal value may depend on localconditions, so it should be an optimization task to refine this setting. All the other channels are then set as a power relative to the CPICH power.

Figure 12 shows an example of power settings in a macro cell with a maximumtransmit power capability of 46 dBm (40 W). The CPICH has been set at 10% (4 W)

of the total power. The primary and secondary SCHs have been set at 6% of totalpower, but they are subject to a 10% duty cycle so they average a combined power of only 0.48 W. The primary and secondary CCPCHs have each been set at 5% (2W), but it is worth noting that there may be multiple SCCPCHs and that the SCCPCHis potentially a variable rate channel. Higher data rates in the SCCPCH would requirea higher power weighting. The PICH and AICH have been set at 1.5% (0.6 W) each,but again it should be noted that this example only shows a single AICH. There is aone-to-one mapping of AICHs to the number of RACH channels configured so therecould be up to eight AICHs on a cell.

The total power allocated to control channels on this example cell is 9.68 W, almost

25% of the total power available on the cell.





Maximum transmit power for the cell is 46 dBm (40 W)

Channel Percentage of

Total PowerPower (dBm)

Power (W)

CPICH 10% 36 dBm 4 W

P&S SCH(inc. 10% duty cycle)

6% (x 2 x 0.1) 33.8 dBm 2.4 W (x 2 x 0.1)

PCCPCH 5% 33 dBm 2 W

SCCPCH 5% 33 dBm 2 W

PICH 1.5% 27.8 dBm 0.6 W

AICH 1.5% 27.8 dBm 0.6 W

Total for control channels 24.2% 39.86 dBm 9.68 W

Figure 12

Downlink Power Weightings

SC2804/S4/v1.1







CPICHHigher weighting for CPICH increasescell radius but reduces capacity

CPICHLower weighting for CPICH reduces cellradius but increases capacity

Figure 13

Varying the CPICH Weighting

SC2804/S4/v1.1





3.4 Utilizing Soft Capacity and Dynamic CPICH Power

The maximum capacity available in a cell is governed in part by the amount of interference that can be tolerated. Yet the level of interference is not simply a productof the load on the cell itself. Interference is also contributed from neighbour cells.Thus a cell’s potential capacity at any moment is partly influenced by the load on itsneighbour cells. This means that a cell can carry more traffic if its neighbours arecarrying less and vice versa.

In real networks the offered traffic is not evenly spread over the ground, thereforeneighbouring cells will tend to carry different loads. A busy cell will be forced totransmit more power in the downlink direction because there are more establishedchannels. This creates more interference to neighbour cells, limiting their potentialcapacity. It would be desirable to balance the load as far as possible between cells inorder to distribute interference more evenly. This should lead to a higher totalcapacity.

Load balancing can be achieved by varying the CPICH weightings among cells.Busier cells would be given lower CPICH weightings to reduce coverage area andload. Quieter cells would have higher CPICH weightings to increase coverage andcapture more offered traffic.

Varying the CPICH weightings can be performed as an optimization function bysetting fixed values based on average conditions. However, traffic characteristics inreal networks are variable, making it hard to find a truly optimal setting. Furthermore,without great care it would be easy to create coverage holes by setting values thatare too low, or to increase the proportion of soft handovers with settings that are toohigh. Either way this would reduce rather than increase overall capacity.

Some vendors may have features that enable the dynamic control CPICH power weighting. This uses an algorithm in the RNC to dynamically adjust power weightingon cells to suit current traffic conditions. The optimizer’s input would then relate tosetting the triggers and constraints for the dynamic weighting control algorithm. They

could, for example, influence whether coverage or capacity is the dominant factor.Simulations of such systems show useful gains in capacity. They also showconsiderable variation in optimal CPICH weightings as high as 60%. This againsuggests that a optimal static value would be hard to find.





Unbalanced load

CPICH Power weightingscan be used to balance load

Figure 14

Soft Capacity

SC2804/S4/v1.1





3.5 Pilot Pollution

Pilot pollution occurs in areas of overlapping coverage between multiple cells.Specifically, it is an area where signal strength is good but there is also a largenumber of non-dominant servers. Signal strength in this case can be considered tobe the CPICH RSCP. In such an area the receiver is not able to decode the downlinkchannels because the multiple good servers interfere with each other to the extentthat the signal-to-interference ratio for the CPICH, Ec/Io, is not good enough despitethe high CPICH RSCP.

In effect, this will result in a coverage hole, where UEs are not able to obtain servicefrom the network. Most 3G planning tools will be able to plot and account for areas of pilot pollution. Nevertheless, planning tools are limited by the accuracy of thepropagation model. This means that a key early optimization task may be to identifyand rectify significant areas of pilot pollution in the built network.

Adjustment of CPICH weightings is one option for dealing with pilot pollution. It canbe used to create a dominant server in affected areas. Other techniques to consider may be antenna adjustments including orientation, downtilt and height.







4.1 Use of Downtilt

In the interests of controlling interference and coverage, it is common to useantennas whose vertical beams are tilted down towards the ground by a fewdegrees.

Looking at Figure 16a, a directional antenna with 0° downtilt is illustrated. For comparison, Figure 16b shows a sector antenna with x° in-built (electrical) downtilt.Note that all lobes are downtilted. Figure 16c shows an antenna with 0º electricaldowntilt which has been mechanically downtilted by x°. Note that the back lobe istilted up and side lobes in the horizontal plane will not be fully downtilted.

Accordingly, electrical downti lting is often preferred, although a combination of electrical and mechanical tilting is common, as shown in Figure 16d.

Another technique involves mechanically uptilting an electrically downtilted antenna,as shown in Figure 16e. This can be used to create an antenna with a heavilydepressed back lobe, which could be useful for interference rejection in some cases.

A typical configuration for UMTS at rollout is to use minimal downtilt to maximizecoverage. As the network matures, downtilt is applied when in-fill cells are built.Ideally, therefore, variable electrical downtilt antennas should be used to facilitate

this. Some estimates are that an antenna’s downtilt could need changing betweenthree and four times in the first five years of operation. This may mean that the mosteconomical solution would be remotely adjustable downtilts. This would greatlyreduce the number of site visits required. It would also allow for the potential dynamicadjustment of downtilts based on load conditions.

Normally, omni antennas can only be tilted by electrical means, as shown in Figures16f and 16g. (Mechanically tilted omni antennas are very rare indeed, but may beseen on steep hillside sites). Typical tilts in use vary from 0 degrees (no tilt) to over 10º. Electrical tilt angles of 0º, 2º, 4º and 8º are common, but others are available.

In general, the majority of tilt should be achieved using electrical tilting, with final fineadjustment being mechanically made.

4 ANTENNA CONFIGURATIONS





Antenna

x°

Sector antenna 0° tilt

Sector antenna x° electrical tilt

Sector antenna x° mechanical tilt

Sector antenna x° electrical tilt y°mechanical tilt

Sector antenna electrically downtilted

mechanically uptilted

Omni antenna 0° tilt

Omni antenna x° electrical tilt

x°

x°

x°

x°

y°

(a)

(b)

(c)

(d)

(e)

(f)

(g)

Figure 16

Antenna Downtilt

SC2804/S4/v1.1





4.2 Calculations for Beamtilt

Figure 17 illustrates how the range from an antenna for the main lobe, as well as theupper –3 dB and lower –3 dB beam extremities, can be calculated. The exampleshown assumes an antenna height of 15 m.

The formulas used are:

Radio horizon = 4.12 h kilometres

Main lobe = h kilometres1000 tan α

Upper –3 dB = h kilometres1000 tan (α –β/2)

Note: if α < β/2 this will be over the horizon

Lower –3 dB = h kilometres1000 tan (β/2 + α)

In all cases:

h = antenna height above average terrain in metresα = downtilt in degreesβ = vertical beam width in degrees





Figure 17

Results for Beamtilt

SC2804/S4/v1.1


ß

– –

downtilt angle

vertical beamwidth

Example:h = 15 metres, vertical beamwidth, ß = 10°, radio horizon = 15.95 kilometres

Horizon

AntennaHeight h

Lower –3 dB

MainLobe

Upper –3 dB

ß

0

1

2

34

5

6

7

8

9

10

11

12

1314

15

15.95

0.85

0.43

0.290.21

0.18

0.14

0.13

0.11

0.10

0.08

0.08

0.06

0.060.06

0.05

over horizon

over horizon

over horizon

over horizonover horizon

over horizon

0.85

0.43

0.29

0.21

0.18

0.14

0.13

0.110.10

0.08

0.18

0.14

0.13

0.110.1

0.08

0.08

0.06

0.06

0.05

0.05

0.05

0.05

0.050.05

0.03

Downtilt ° Main Lobe (km) Lower –3 dB (km) Upper –3 dB (km)




4.3 Practical Antenna Types and Tilt Effects

It is important when considering downtilts to be aware of the effects likely to beproduced in realistic environments. The specific effect will depend on the type of antenna and on the antenna characteristics.

It is common for an operator to use a mix of vendors for similar types of antenna;often this is driven by cost or availability at the time of site build. However, theoptimizer should be aware that specific characteristics, even for the same type of antenna, will differ with different vendors. These differences could be significantwhen making changes in UMTS. Thus any simulations performed to test the likelyeffects of a change should always be done with the correct vendor’s antenna data.

The different characteristics of mechanical and electrical downtilt are illustrated inFigures 19a to 19d. Figure 18 shows the terrain and basic site characteristics for thesimulations.





Figure 18

Antenna Characteristics Simulation Parameters

SC2804/S4/v1.1


Antenna at 15m

PA Max 37 dBm




4.3.1 Changing Antenna Type

Figures 19a to 19d show the differences between several common antenna types. All the antennas shown are designed for use with UMTS.

Figure 19a is an omnidirectional antenna from Cellwave. It has a maximum gain of 7.65 dBi. Note the impact that terrain has on the omnidirectional radiation pattern.

Figure 19b shows a 85º directional antenna from CSA Wireless. It has a maximumgain of 16 dBi, a horizontal beam width of 85º and a vertical beam width of 7º. It isfitted with 2º of electrical downtilt and is mounted with 0º of mechanical downtilt.

Figure 19c shows a 65º directional antenna from CSA Wireless. It has a maximumgain of 17.5 dBi, a horizontal beam width of 65º and a vertical beam width of 9º. It isfitted with 2º of electrical downtilt and is mounted with 0º of mechanical downtilt.

Figure 19d shows a 33º directional antenna from CSA Wireless. It has a maximumgain of 20 dBi, a horizontal beam width of 33º and a vertical beam width of 6.5º. It isfitted with 2º of electrical downtilt and is mounted with 0º of mechanical downtilt.












Figure 19d

33 20 2elec 0mech

Figure 19c

65 17 2elec 0mech




4.3.2 Applying Mechanical Downtilt

Figures 20a to 20f show the effect of applying mechanical downtilt up to 16º for the65º directional antenna.

Figure 20a shows 0º mechanical downtilt for the 65º antenna.

Figure 20b shows 4º mechanical downtilt for the 65º antenna.

Figure 20c shows 8º mechanical downtilt for the 65º antenna.

Figure 20dshows 12º mechanical downtilt for the 65º antenna.

Figure 20e shows 16º mechanical downtilt for the 65º antenna.

Figure 20f shows the vertical and horizontal radiation patterns for the 65º antenna.Note the upper side lobes in the vertical radiation pattern and compare these with thecoverage predictions






Figure 20b

65 17 2elec 4mech

Figure 20a

65 17 2elec 0mech





Figure 20d

65 17 2elec 12mech

Figure 20c

65 17 2elec 8mech






4.3.3 Applying Electrical Downtilt

Figures 21a to 21c show the effect of applying electrical downtilt up to 6º for the 65vdirectional antenna.

Figure 21a shows 0º electrical downtilt for the 65º antenna.

Figure 21b shows 4º electrical downtilt for the 65º antenna.

Figure 21c shows 8º electrical downtilt for the 65º antenna.

Note that electrical downtilt applies in all azimuths whereas the mechanical downtiltapplies predominantly in the bore sight of the antenna. Combinations of electricaland mechanical downtilt can be used to shape the coverage pattern with great effect.











Figure 21c

65 17 6elec 0mech

SC2804/S4/v1.1







Low visual impact

Low cost

Low wind loading

Low maintenance

Rapid fitting

Different azimuths

Different downtilts

Separately optimized

Separately upgraded

Consideration Type of UMTS Antenna

Separate Integrated/Multiband

(Broadband)

Figure 22

Choice of Antennas

SC2804/S4/v1.1





5.1 Minimum Coupling Loss

Maintaining acceptable noise levels at the Node B receiver for UMTS is veryimportant if uplink capacity and coverage are to be maximized. In general this is thefunction of the closed loop power control process. As the UE approaches the Node Bthe power control process steps its power down. The UE is required1 to be able tostep its power down to at least –50 dBm.

Figure 23 shows a situation where the mobile’s power has been stepped down to –50 dBm. The service supported is 12.2 kbit/s speech. The Node B requirement2 for sensitivity level at this service bit rate is –121 dBm. This figure is quoted for staticconditions and includes the processing gain and the receiver noise figure.

Processing gain at 12.2 kbit/s is 25 dB, a required Eb/No of 5 dB is assumed and a 3dB interference margin has been allowed for other traffic on the cell. The result isthat uplink power control should be aiming for –104 dBm at the Node B. This gives acoupling loss (including antenna gains) of 54 dB. At this value of coupling loss thesignal level is just sufficient to meet the bit error requirements in the channel. If theUE were to move closer in order to reduce the coupling loss it would be contributingmore power at the receiver. Since its power cannot be reduced further it will have theeffect of reducing capacity. Thus, in this example, 54 dB is the Minimum CouplingLoss (MCL). This is a fairly typical value.

Assuming a typical combined antenna gain figure for the UE and the Node B of 16dBi, the minimum path loss will be about 70 dB. This is only likely to occur with clear line of site over very short distances. For free space this would mean a distance of less than about 40 m.

1 3GPP TS 25.104 User Equipment (UE) Radio transmission and reception (FDD).2 3GPP TS 25.104 BS Radio transmission and reception (FDD).

5 RADIO PERFORMANCE





–50 dBm

Coupling Loss

–104 dBm

–121 dBm

–129 dBm

Coupling Loss

= 54 dB

Eb/No 5 dBIM 3 dB

Gp 25 dB

Figure 23

Minimum Coupling Loss

SC2804/S4/v1.1





5.2 Adjacent Channel Leakage Ratio (ACLR)

Limitations in design capability mean that a transmitter will always radiate somepower outside the intended radio channel. Since there is very little space betweenadjacent channels in UMTS, and since UMTS is sensitive to interference, the Adjacent Channel Leakage Ratio (ACLR) is an important consideration.

The specified performance requirements for ACLR are shown in Figure 24 for the UEand for the Node B (both in FDD mode).





–33 dB –33 dB

–43 dB –43 dB

fc – 10 MHz fc – 5 MHz fc fc + 5 MHz fc + 10 MHz

UE ACLR Performance Requirement for FDD Mode

–45 dB –45 dB

–55 dB –55 dB

fc – 10 MHz fc – 5 MHz fc fc + 5 MHz fc + 10 MHz

Node B ACLR Performance Requirement for FDD Mode

Figure 24

Adjacent Channel Leakage Ratio (ACLR)

SC2804/S4/v1.1





5.3 Radio Carrier Spacing

Nominal channel spacing for UMTS radio carriers is 5 MHz, but the standards allowfor the centre frequencies to be altered in 200 kHz steps. Each 200 kHz step isdescribed with a UMTS Absolute Radio Frequency Channel Number (UARFCN). Anoperator may take advantage of this by creating larger guard bands betweenadjacent radio carriers.

This is most likely to be the case where adjacent radio carriers belong to differentoperators, as shown in Figure 25. Nevertheless, there is no specified or technicalrestriction demanding that an operator uses consistent UARFCNs across their network. Thus, for example, an operator with space for three radio carriers couldallow much larger guard bands in geographical areas where only two radio carriersare in use. This would increase the potential capacity of each of the individual radiocarriers.





Operator A

Enhancedguard band

Operator B

5.4 MHz 4.8 MHz 4.8 MHz

Operator A

Enhancedguard bands

Operator B

Figure 25

Radio Carrier Spacing

SC2804/S4/v1.1





5.4 Adjacent Channel Interference (ACI)

There are a number of scenarios where adjacent channel interference may have asignificant bearing on system performance. The example illustrated in Figure 26involves interference between two cell layers. Node B 1 is a micro cell and Node B 2is a macro cell. Node B 2 is the serving cell for the UE. In this scenario the UE is veryclose to the micro cell, but on the edge of coverage for its serving cell. The result isthat it is transmitting a large amount of power on an adjacent channel very close tothe micro cell’s receiver.

5.4.1 Worst Case Assumption

Consider the worst case for the scenario illustrated in Figure 26.

If the UE is class 4 it may be transmitting 21 dBm assuming a minimum coupling lossto the micro cell of 54 dB and ACLR performance for the UE in the adjacent channelof 33 dB. The interfering signal level at the input to the micro cell receiver will be:

ACI level = 21 – 54 – 33= –66 dBm

Clearly, an interfering signal level signal level of –66 dBm will have a serious impacton the performance of the micro cell. However, it is important to appreciate that thisis a worst-case scenario, and if the two cells belong to the same operator thesituation should be avoidable. In this case a hard handover from the macro to themicro cell would seem appropriate. This situation would be most problematic whenthe cells belonged to different operators.









Options to reduce inter-operator ACI:

revise antenna position

desensitize receiver

revise UARFCNinter-operator cooperation

Inter-operator ACI

Operator A

Micro cellUE

Operator B

Operator BMacro cell

Figure 27

Inter-Operator ACI

SC2804/S4/v1.1







GSMBTS

UMTSNode B

Antenna SystemGSM UMTS

Isolation = 30 dB

–96 dBm –126 dBm

GSMBTS

UMTSNode B


Isolation = 30 dB

–128 dBm –98 dBm

Figure 28

Emission Limits for Current Equipment

SC2804/S4/v1.1





6.1.2 Older GSM Equipment

Co-locating a UMTS Node B and older GSM 900 or especially GSM 1800 equipmentrequires additional isolation measures. This is because older GSM equipment wasnot required to suppress out-of-band spurious emissions to the current degree. In theUMTS band, it is required only that spurious emissions be kept to less than –30dBm. Unless additional isolation is provided, the normal 30 dB of antenna isolationwould result in a –60 dBm input to the Node B, with disastrous desensitizationresulting. The effective uplink range of the Node B would be severely reduced.

Because at least 60 dB are required, it may be impractical to increase the isolationby antenna spacing alone. One solution would be to include an in-line bandpass filter with a steep roll-off characteristic in the output from the GSM BTS. Such filters willintroduce a small (1 to 2 dB) additional downlink loss for the GSM cell. This willcause a small reduction in range unless more BTS output power can be obtainedfrom the TRXs to maintain the Effective Isotropic Radiated Power (EIRP). Thisassumes the filter is placed in the GSM transmit branch, i.e. between the combiner and duplex filter (if fitted).





GSMBTS

UMTSNode B


Isolation = 30 dB

–30 dBm –60 dBm

BandpassFilter

UMTSNode B


Isolation = 30 dB

–30 dBm <–120 dBm

older equipment

GSMBTS

GSM UMTS

Node Bdesensitized

older equipment Node B notdesensitized

filter includedonly in GSMtransmit branch

> 60 dB

attenuation inUMTS band

Figure 29

Emission Limits for Older Equipment

SC2804/S4/v1.1







GSM 900BTS

UMTSNode B

Antenna SystemIsolation = 30 dB

GSM 900BTS

UMTSNode B

Antenna SystemIsolation = 30 dB

8 dBm 43 dBm

43 dBm 8 dBm

Feeder/Connector loss = 2 dB




Figure 30

Receiver Blocking – UMTS and GSM 900

SC2804/S4/v1.1













SECTION 5

IDLE MODE AND SYSTEM ACCESS

i









1 PLMN and Cell Selection 5.11.1 Selection and Idle Mode Activities 5.11.2 PLMN Selection 5.31.3 Capturing Roaming Subscribers 5.51.4 Cell Selection 5.7

2 Exercise 1 – Cell Selection Scenarios 5.13

3 The Random Access Channel (RACH) 5.153.1 Applications for RACH 5.153.2 RACH Operation 5.173.3 RACH Control Parameters 5.193.4 RACH Optimization Considerations 5.21

4 Cell Reselection 5.234.1 Introduction 5.234.2 Basic Cell Reselection Process 5.23

4.3 Basic Inter-RAT Reselection 5.274.4 Reselection with Hierarchical Cell Structures (HCS) 5.294.5 Inter-RAT Reselection with HCS 5.39

5 Exercise 2 – Cell Reselection Scenarios 5.41

6 Radio Resource Control (RRC) Functions 5.436.1 Introduction 5.436.2 Cell Access Restrictions 5.456.3 Admission Control 5.49

CONTENTS

iii










• describe the UMTS cell selection process

• characterize the effect of each parameter relating to cell selection

• describe the UMTS cell reselection process

• characterize the effect of each parameter relating to cell reselection

• describe the configurations and capabilities of the RACH

• describe the open loop power control process and controlling parameters

• analyze the effect of changing open loop power control parameters

• describe the interactions between UMTS and GSM/GPRS in respect of cell

selection and reselection

• describe the functions of RRC in terms of admission control

• explain how traffic prioritization and mapping to appropriate channel types

can be used to optimize capacity in the radio access network

OBJECTIVES

v





1.1 Selection and Idle Mode Activities

At switch-on the UE has a number of tasks to perform to ensure that it is in acondition to obtain services through a network as required. The first of these is toperform PLMN selection. The selection process is performed by the NAS part of theprotocol stack, and may involve input from the user.

Having selected a Public Land Mobile Network (PLMN) the UE is required to selectand camp-on a suitable cell belonging to the selected PLMN. Registration is thenperformed through the camped-on cell. After a successful registration the UE willassume the camped normal state and begin idle mode tasks.

Idle mode tasks will involve neighbour cell measurements, cell reselection, systeminformation monitoring and paging monitoring. The precise behaviour of the UE whenperforming these tasks will depend upon the camped on cell’s channel configurationand the setting of several related parameters in system information. These actionsare fully defined in the UMTS standards.1

1 3GPP TS 25.304 UE procedures in idle mode and procedures for cell reselection in connected mode.3GPP TS 23.122 NAS functions related to MS in idle mode.

1 PLMN AND CELL SELECTION









AvailablePLMNs

SelectedPLMN and

RAT

SIM

Readpriorities

NAS

Builds priority list from available PLMNsand selects a PLMN and RAT usingeither automatic or manual mode:

automaticmanual

highest priorityuser selected

AS

Scans for and measures availablePLMNs. Supplies a list of all PLMNssuccessfully identified to NAS. Mayuse stored information to optimize theprocess.

user PLMN listoperator PLMN list

Figure 2

PLMN Selection

SC2804/S5/v1.1







Priority List

1 HPLMN

2 User-defined priority from SIM

3 Operator-defined priority from SIM

4 PLMNs meeting high-quality criterion in

random order

5 Other PLMNs in order of signal quality

Assumed not available

Likely to be empty unless asubscriber has a strong preference

Driven by commercial

relationships between operators

Best strategy is to providea good signal level

Figure 3

Optimizing PLMN Selection

SC2804/S5/v1.1





1.4 Cell Selection

Following PLMN selection, the AS will be provided with the selected PLMN and theradio access technology to use. It is possible that the selected radio accesstechnology will be GSM, in which case the cell selection process will be as describedin GSM specifications.1

Assuming the radio access technology to be used on the selected PLMN is UMTS,the UE will find the strongest cell on each carrier and test it for suitability. The UE willselect the first suitable cell it finds. A suitable cell is defined in the following way:

• it belongs to the selected PLMN or an equivalent PLMN• it is not barred

• it is not in a forbidden location area for roaming

• the cell selection criteria are fulfilled

Equivalent PLMNs will be indicated in system information and their use allowsinfrastructure sharing between operators. Cell bar status is also indicated in systeminformation. The use of forbidden location areas for roaming allows operators toprovide national roaming on a regional basis.

The cell selection criteria are shown in Figure 4. As indicated, a UMTS FDD cell isconsidered suitable if both Srxlev and Squal are greater than zero. The systemparameters used in this calculation are broadcast in system information and they areas follows:

• Qqualmin (–24 to 0 dB)

• Qrxlevmin (–115 to –25 dBm in 2 dB steps)

• UE_TXPWR_MAX_RACH (–50 to 33 dBm)

Qqualmin and Qrxlevmin are minimum required levels for the cell.UE_TXPWR_MAX_RACH is the maximum power that a UE is allowed to use on theRACH in the cell. Potentially these parameters could be adjusted by an optimizer toinfluence a UE’s behaviour. However, the most likely way to influence cell selectionwould be through adjustment of radio characteristics such as transmit power or antenna tilt.

The other parameters used in the calculation are Qqualmeas and Qrxlevmeas, which are aUE’s measured values of CPICH Ec/No and CPICH RSCP respectively for a cell. Additionally, P_MAX is the UE’s maximum transmit power capability.

1 3GPP TS 03.22 Functions related to mobile station in idle mode and group receive mode.







1.4.1 Initial and Stored Cell Selection

The UE may use either initial or stored cell selection. For initial cell selection the UEwill have no information on frequencies or scrambling codes used by the selectedPLMN. For stored cell selection, the more typical case, the UE will have storedpreviously received information elements relating to frequencies and scramblingcodes used, which may speed up the process.





A cell is suitable if:

it belongs to the selected PLMN or an equivalent PLMNit is not barredit is not in a forbidden location area for roamingthe cell selection criteria are fulfilled

The cell selection criteria for UMTS FDD cells are fulfilled when Srxlev and Squal areboth greater than zero.

Where:

Squal = Qqualmeas – Qqualmin

Srxlev = Qrxlevmeas – Qrxlevmin – Pcompensation

and

Pcompensation = max(UE_TXPWR_MAX_RACH – P_MAX, 0)

Figure 4 (repeated)

Cell Selection

SC2804/S5/v1.1







Qqualmin = –12 dB

Qrxlevmin = –97 dBm

UE_TXPWR_MAX_RACH = 24 dBm

Power Class 3 = 24 dBm

Qqualmeas = –10 dB

Qrxlevmeas = –90 dBm

= Qqualmeas –Qqualmin

= –10 ––12

= 2

= Qrxlevmeas –Qrxlevmin –Pcompensation

= –90 ––97 –max(24 –24, 0)

= 7 –0

= 7

Squal

Srxlev

Squal and Srxlev are both greater than zero, therefore the cell is selected.

Figure 5

Cell Selection Criteria Example

SC2804/S5/v1.1





Figure 6 shows a cell and the associated broadcast parameter values for Qqualmin,

Qrxlevmin and UE_TXPWR_MAX_RACH. It also shows values for Qqualmeas andQrxlevmeas for three positions in the cell, A, B and C.

1 For positions A, B and C calculate whether the cell selection criteria are fulfilledfor a power class 4 UE (21 dBm).

2 For positions A, B and C calculate whether the cell selection criteria are fulfilledfor a power class 3 UE (24 dBm).

3 What might you adjust to ensure that the cell appears suitable for both types of UE in all locations?

2 EXERCISE 1 – CELL SELECTION SCENARIOS





Qqualmin = –12 dBQrxlevmin = –95 dBm

UE_TXPWR_MAX_RACH = 24 dBm

C

Qqualmeas = –11 dBQrxlevmeas = –93 dBm

BQqualmeas = –10 dB


AQqualmeas = –9 dB


Figure 6

Exercise 1 – Cell Selection Scenarios

SC2804/S5/v1.1





3.1 Applications for RACH

Having selected a cell, the UE will have to perform a registration. This will involve alocation update to the circuit-switched core network and a routing area update to thepacket-switched core network. This may be performed as a combined or as separateprocedures. In order to perform these signalling procedures an RRC connectionmust be established. This will involve at least initial access using the RACH, butoptionally the whole procedure could be performed on the RACH.

The RACH is an uplink-only contention-based channel utilizing open loop power control. Its use is mandatory for initial access, but it has several other optionalfunctions. Here, initial access prior to a registration procedure is being described, butinitial access is also required to initiate packet data activity while RRC connected inthe CELL_FACH, CELL_PCH or URA_PCH sub-states. The RACH may optionallybe used for ongoing exchange of signalling or packet data; for example, a completeexchange of all messages in a location update, or the transfer of small packet datasuch as SMS or telemetry information. Thus the RACH is a multipurpose channelwhose activity rate is not limited to the initiation of signalling procedures.

Given the significant amount of RACH activity that can be expected in a UMTS cell itis important that the parameters that control its operation are set with care. In therollout phase, activity will be limited and therefore it should be acceptable to use the

same parameter values on all cells. However, inappropriate values may result inexcessive interference from RACH channel usage as traffic load increases.Therefore, as a system matures and cell traffic load rises, it will be necessary tooptimize parameters on the most affected cells.

3 THE RANDOM ACCESS CHANNEL (RACH)





Figure 7

Applications of RACH

SC2804/S5/v1.1





3.2 RACH Operation

The RACH is a transport channel and as such is mapped onto a physical channelcalled the PRACH. It is the operation of the physical channel that is of most interestto the optimizer.

The principal concern is the setting of parameters that control the open loop power control required in this contention-based channel. Figure 8 illustrates the basicoperation of the physical channel that carries the RACH. As shown, the timingstructure for the RACH occupies frames of 20 ms. These are divided into 15 accessslots, each with duration 5,120 chip periods. There may be up to 16 RACH channelsavailable on a cell. The UE will determine from system information the configurationof RACH on the cell and if there are any applicable access restrictions to channels or access slots within channels.

Assuming the UE has appropriate access rights it begins the procedure bycalculating an initial power. It then randomly selects one of 16 signature codes andtransmits this in a preamble part of duration 4,096 chip periods. It will then monitor the AICH associated with the RACH. The AICH can be used to provide either apositive or a negative response to the preamble. A negative response would causethe UE to abandon the procedure. However, it is most likely (and desirable) that after the first preamble there will be no response in the AICH. In this case, either three or

four access slots after the start of the first preamble, the UE will transmit a secondpreamble at a slightly higher power and again monitor the AICH. The power stepused and the delay before retransmission are parameters read from systeminformation. If the UE still has no response the pattern will be repeated; a thirdtransmission with another power step up. This will continue until the UE gets aresponse in the AICH or until it has reached the maximum allowable number of preambles.

Assuming that the UE does get a positive response in the AICH, it will begintransmission of the message part. This starts either three or four access slots after the start of the successful preamble and is transmitted at the same power as the

successful preamble (or with a parameter defined offset). The duration of themessage part may be either 10 ms or 20 ms.





10 ms Radio Frame

20 ms Access Frame

DLAICH

Initialpower

ULRACH

Power step

Power step

Firstpreamble

Secondpreamble

Thirdpreamble

Noresponse

No

response

Positive

response

Message Part 10 ms or 20 ms

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14

Figure 8

RACH Basic Operation

SC2804/S5/v1.1





3.3 RACH Control Parameters

There are many parameters that describe RACH configuration and access rules thatmust be read from system information1 by the UE. Among other things theseparameters will describe the number of RACH channels configured, the bit ratesavailable, access class applicability to channels and slots, signatures available andscrambling codes to be used.

The parameters most applicable to the radio optimizer are those relating to openloop power control and the access procedure itself. These parameters are listed inFigure 9.

An important value is the transmit power used for the first preamble transmission.The UE uses the formula shown2 to calculate the initial power. Primary CPICH TXpower, UL Interference and Constant value are all found in system information. Thevalue of CPICH RSCP is measured by the UE.

Other parameters include the Power Ramp Step, Preamble retrans max and Mmax.This final parameter determines the maximum number of preamble cycles in a RACHattempt, whereas Preamble retrans max determines the maximum number of preambles in a cycle. NB01min and NB01max are limits for a random backoff timebetween preamble cycles.

1 3GPP TS 25.331 Radio Resource Control (RRC) protocol specification.

2 3GPP TS 25.331 Radio Resource Control (RRC) protocol specification.





Time

Backoff Backoff Preamblecycle

Preamble_Initial_Power = Primary CPICH TX power –CPICH_RSCP –UL Interference + Constant value

Power Ramp Step

1 to 8

Preamble retrans max

Mmax

NB01minNB01max

1 to 64

1 to 32

0 to 50

–110 to –70 dBm –35 to 10 dB

–115 to –25 dBm –10 to 50 dBm

Figure 9

RACH Power and Access Control Parameters

SC2804/S5/v1.1





3.4 RACH Optimization Considerations

Optimization activity for the RACH is likely to be focused on the operation of openloop power control. Since interference is a capacity-limiting factor in a UMTS cell it isdesirable to limit the interference contribution from RACH activity as much aspossible. Ultimately the aim is to set parameters that enable UEs to find the correctpower level quickly and reliably from any location in the cell.

The key parameter in this respect is the Constant value, which is used to calculateinitial power. If initial power is too high, mobiles’ RACH transmissions will causesignificant interference to other traffic and signalling channels in the cell. This in turnwill lead to higher average power being set by closed loop power control processesand a resulting loss in capacity. If initial power is too low, mobiles may need totransmit a large number of preambles before reaching a successful power level. Inextreme cases there may also be a large number of aborted RACH attempts. Thecumulative noise from a larger number of preamble transmissions will also result inhigher levels of interference and in degraded capacity.

An ideal setting would result in the first preamble failing to get a response from thecell with a positive response subsequently being returned in the AICH after thesecond preamble. An appropriate value for this condition will depend on radioconditions in the cell and the amount of activity required on the RACH channel.

The number of preambles transmitted before a positive response is seen in the AICHis a good indicator of performance. This could be observed with an appropriate drivetest tool used in the cell of interest. It should be set to make a large number of short-duration calls. If the first preamble consistently gets a response then the initial power is probably too high. In this case a decrease in the parameter Constant value shouldbe considered. The opposite would be true for a consistently large number of preambles before a positive response is received. It is important that anyassessment of the number of preamble attempts is averaged across a representativegeographical area for the cell.





Required power

Time

Single preamble successfulInitial power too high

Decrease Constant value

Required power

Time

Multiple preambles requiredInitial power too low

Increase Constant value

Required power

Time

Ideal

Figure 10

Setting the Constant Value

SC2804/S5/v1.1





4.1 Introduction

Once a UE is registered with a PLMN on a suitable cell it will assume idle mode andbegin a monitoring process that may lead to cell reselection. This in turn may resultin a location update, a routing area update, an UTRAN registration area update or acell update depending on RRC connection status.

This is because the UMTS idle mode cell reselection process is also used in thecontext of connected mode for certain types of packet data or signalling transactions.This occurs when the UE is in the CELL_FACH, CELL_PCH or URA_PCH states.

There are many options for the way in which cell reselection is managed in UMTS.The mechanism used will depend on the architecture of the network and operator preferences.

4.2 Basic Cell Reselection Process

At rollout a UMTS network is likely to contain only a single cell layer. All UMTS cellswould therefore be on the same frequency. This simplifies the cell reselectionprocess in two ways. Firstly, the UE only needs to make intra-frequency UMTSmeasurements; and secondly there is no need to consider the hierarchical the cell

reselection criteria. Although many networks will need to support inter-RATreselection at rollout, typically to and from GSM, it is possible to manage this withoutuse of the hierarchical cell structures.

4.2.1 Basic Measurement Rules

It is possible to limit the amount of neighbour cell measurement performed by the UEwhen the service from the current serving cell is adequate. This is controlled bysetting the parameters Sintrasearch, Sintersearch and SsearchRATm. For UMTS FDD theseparameters are applied by the UE, as shown in Figure 11. These are optional

parameters and if they are not included in system information the UE will performmeasurements on all indicated neighbour cells irrespective of the condition of theserving cell.

4 CELL RESELECTION





Sintrasearch

(–32* to 20 in steps of 2)

Sintersearch(–32* to 20 in steps of 2)

SsearchRATm

(–32* to 20 in steps of 2, value 20 isinterpreted as absent)

* Negative values are considered tobe 0.

Only perform intra-frequency measurements if:

Squal • Sintrasearch

Only perform inter-frequency measurements if:

Squal • Sintersearch

Only perform inter-RAT measurements if:

Squal • SsearchRATm

Qqualmeas

(–25 to 0) Calculate CompareSqual

(1 to 25)

UEServing Cell(UMTS FDD)

Figure 11

Basic Measurement Rules

SC2804/S5/v1.1





4.2.2 Basic Cell Reselection Criteria

The UE tests all measured cells, including the current serving cell, against the cellselection criteria. All cells that meet the cell selection criteria are then ranked usingthe cell-ranking criterion ‘R’.

A new cell will be selected if it is ranked higher than the current cell for a time intervaldefined by the parameter Treselection, and if the UE has been camped on thecurrent serving cell for more than one second.

Note that the values of Qqualmin, Qrxlevmin and UE_TXPWR_MAX_RACH used tocalculate the cell selection criterion S can differ for each neighbour cell. This couldbe used to influence whether a neighbour cell is judged good enough to consider for reselection. However, these values apply to the neighbour cell for all cell selectioncircumstances, so should not be changed when trying to optimize only onereselection scenario.

The key parameters for optimizing the cell reselection process for a single scenarioare as used to calculate the ranking criterion ‘R’. The way that these parameters areapplied is shown in Figure 12. (Note, Figure 12 only covers the case where allneighbour cells are UMTS FDD).

The parameters used in the cell ranking criterion and reselection are as follows:

• Rs –calculated ranking value for the serving cell

• Rn –calculated ranking value for a neighbour cell

• Qmeas,s –can be set as either Q qualmeas or Qrxlevmeas (serving cell)

• Qmeas,n –can be set as either Q qualmeas or Qrxlevmeas (neighbour cell)

• Qhysts –hysteresis value, 0 to 40 dB is steps of 2

• Qoffsets,n –offset value, –50 to 50

• Treselection –timer value, 0 to 31 seconds

Note that a single value of Qhysts is set for the serving cell, but the value of Qoffsets,n

can be set independently for each listed neighbour cell.





Neighbour Cell Measurements(Based on measurement rules)

UE applies cell selection criterion ‘S’

Squal > 0 and Srxlev > 0 where:

Squal = Qqualmeas –Qqualmin

Srxlev = Qrxlevmeas –Qrxlevmin –Pcompensation

Cells meeting the ‘S’ criterion are ranked

using the ranking criterion ‘R’

Rs = Qmeas,s + Qhysts

Rn = Qmeas,n + Qoffsets,n

A neighbour cell is reselected if:

it is ranked higher than the serving cell for a timegreater than Treselection

the UE has been camped on the current servingcell for at least one second

Figure 12

Basic Cell Reselection Criterion

SC2804/S5/v1.1





4.3 Basic Inter-RAT Reselection

Even without using reselection procedures for HCS it is possible to allow reselectionof inter-RAT cells from UMTS FDD. There are two main cases for this, reselection toa GSM cell and reselection to a UMTS TDD cell.

4.3.1 Reselection to GSM

GSM neighbour cell information is included in system information. The downlinkmeasurement used for assessment of a GSM cell is RSSI indicated in dB. There isno specific quality measure for a GSM cell, so the cell selection criterion is met for aGSM cell if Srxlev is greater than or equal to zero.

Appropriate setting of SsearchRATm can also be used to control reselection to GSMsince it can prevent GSM cells being considered when not required.

The value of Qrxlevmin, used to calculate Srxlev, is set for individual cells. Thereforefor a GSM cell it can be set to provide a weighting for or against performing an inter-RAT reselection. Similarly, the value MS_TXPWR_MAX is used to calculatePcompensation, which could also be used to influence the likelihood of an inter-RATreselection.

Assuming neighbour cell measurements are performed on a GSM cell and that itmeets the cell selection criterion, it must be ranked before it can be reselected.When a GSM cell is ranked the value of Qmeas,n will also be RSSI. Again, the value of Qoffsets,n is set individually for each neighbour cell and can be used to weight thelikelihood of a GSM cell being reselected.

4.3.2 UMTS TDD Reselection

For UMTS TDD mode the cell selection criterion ‘S’ is based only on Srxlev being

greater than or equal to zero. As for GSM, appropriate values of maximum transmitpower and Qrxlevmin can be set to influence the value of Srxlev. This may preventthe TDD cell even being considered for reselection.

In respect of ranking, the value of Qmeas,n will be P-CCPCH RSCP and anappropriate value of Qoffsets,n can be set to weight the likelihood of reselection.





Neighbour Cell(GSM)

Serving Cell(UMTS FDD)

SsearchRATm –used to control ne ighbour cell measurements

Qrxlevmin and Pcompensation –used to control neighbour cellmeasurements and consideration for ranking

Qoffsets,n –used to control ranking of the considered neighbour cell

Measurements:

Qqualmeas

Qrxlevmeas

Measurement:RSSI

UE uses RSSI

for Qrxlevmeas

and Qmeas,n

Figure 13

Reselection to GSM without HCS

SC2804/S5/v1.1





4.4 Reselection with Hierarchical Cell Structures (HCS)

The UMTS standards1 allow for the cell reselection procedures to be modified toaccount for the use of hierarchical cell structures. This is done in such a way thatreselection depends on the relationship between the cell layer to which the servingand neighbour cell belong as well as the mobility of the UE.

Whether reselection procedures with HCS are to be used or not is indicated insystem information by the parameter Use of HCS.

4.4.1 HCS Priority Levels

In more mature UMTS systems it is likely that an operator will use a hierarchicalarchitecture. Typically this may involve macro, micro and pico cells.

Each HCS layer is allocated a priority level. The standards allow for up to eightpriority levels numbered 0 to 7. HCS priority level 0 is lowest and HCS priority level 7is highest. The highest priority level should be allocated to the smallest or theoverlaid cells. Thus the example shown in Figure 14 has pico cells allocated level 7,micro cells allocated level 6 and macro cells allocated level 5.

Inter-RAT cells can also be treated as HCS cell layers. Therefore GSM or TDD cellswould also be allocated an HCS priority level. It is likely that this would be a lower priority than UMTS FDD macro cells.

1 3GPP TS 25.304 UE procedures in idle mode and procedures for cell reselection in connected mode.





HCS Level 5

HCS Level 6HCS Level 6

HCS Level 6

HCS Level 5

HCS Level 7

HCS Level 7

HCS Level 7

HCS Level 7

Level 5Level 6Level 7

–Macro cell –Micro cell –Pico cell

Figure 14

HCS Priority Levels

SC2804/S5/v1.1





4.4.2 HCS Measurement Rules for UMTS

As can be seen in Figure 15, measurement rules are changed for HCS intra- andinter-frequency measurements. Measurements are only performed on all listedneighbour cells if both if both the standard S intersearch threshold and the additionalHCS search threshold are passed. The threshold for HCS is called SsearchHCS. Itcan be assigned values in the range –105 to 91 dB in steps of 2 and specifies aminimum value of Srxlev before all neighbour cells are measured.

If the two thresholds SsearchHCS and Sintersearch have not been passed, and also if theSintrasearch threshold has not been passed, then the UE will only measure cells with ahigher HCS priority level. For example if the UE was camped on a macro cell it mightonly be measuring neighbour cells that were micro or pico cells.

If, in the above case, Sintrasearch is passed then the UE measures cells with an equalas well as those with a higher HCS priority level.





If

Squal • Sintersearch HCS

The UE measures all listed neighbour cells

or if

Squal > Sintersearch or Srxlev > SsearchHCS

and

Squal > Sintrasearch

The UE measures only cells with a higher HCS priority

than the current serving cell

or if

Squal > Sintersearch or Srxlev > SsearchHCS

and

Squal Sintrasearch

The UE measures only cells with a equal or higher HCS prioritythan the current serving cell

Figure 15

HCS Measurement Rules for UMTS

SC2804/S5/v1.1





4.4.3 Consideration of UE Mobility

Normal HCS reselection rules assume the UE is in a low-mobility state. If the UE is ina high-mobility state then the last two mentioned conditions are overruled. This stateis determined using the parameters TCRmax and NCR. These two parameters are usedto consider the UE’s recent history in terms of the number of reselections performedin a given time period (TCRmax). If the number of reselections performed in this periodexceeds NCR then the UE considers itself in a high-mobility state.

If the UE is in a high-mobility state then unless both the Sintersearch and the SsearchHCS

have been passed it will measure neighbour cells with an equal or lower HCS prioritylevel.

The example shown in Figure 16 illustrates this. Consider that the UE is currentlycamped on a micro cell and that neither the SsearchHCS or the Sintersearch thresholdshave been passed. If the UE was in a low-mobility state then it would only measurecells of equal or higher HCS priority level: in this case other micro or pico cells.However, if the UE was fast moving, perhaps in a vehicle, it is likely that there wouldhave been several recent cell reselections. In this case the UE may have passed thethreshold for high mobility. If the UE is in this state it will only measure cells of equalor lower HCS priority level.





7

7

7566

UE B

UE is high-mobility if

more than NCR reselections have

happened in the lastTCRmax seconds

Mobile A –low mobility,

measures onlyhigher HCRpriority cells

Mobile B –high mobility,measures onlylower HCR prioritycells

7

7

6

5

UE A

Figure 16

Consideration of UE Mobility

SC2804/S5/v1.1





4.4.4 Cell Reselection Criteria with HCS

The UE tests all measured cells, including the current serving cell, against the cellselection criterion. All those cells that meet the cell selection criterion are then further tested against the ‘H’ criterion. This criterion is used to determine if HCS cellreselection rules are to be applied. If there are cells that meet the ‘H’ criterion thenfrom that set only those with the highest HCS priority level will be ranked for reselection. If no cells meet the ‘H’ criterion then all cells meeting the ‘S’ criterion areranked as if HCS did not apply.

As shown in Figure 17 the ‘H’ criterion uses the parameters Qhcss, Qhcsn, TOn andL

n

. Qhcss/n

is specified in system information for each cell and is the quality thresholdfor applying HCS prioritized ranking. The value range for Qhcss/n depends on thetype of cell being considered: for UMTS FDD it is either –25 to 0 or –115 to –26(dependent on quality measure), for UMTS TDD it is –115 to –26, and for GSM it is –110 to –37.

TOn is a temporary offset used to control short-duration reselections. It is setindividually for each cell and can have the value 3, 6, 9, 12, 15, 18, 21 or infinite. TOn

is applied for a time set by the parameter PENALTY_TIMEn, which may have thevalue 0, 10, 20, 30, 40, 50 or 60 seconds. The parameter Ln may take the value 0 or 1 and is dependent on the HCS priority of the cell being considered. It is set to 0 if

the neighbour cell’s HCS priority is the same as that of the serving cell, otherwise it isset to 1.

A new cell will be selected if it is ranked higher than the current cell for a time intervaldefined by the parameter Treselection, and if the UE has been camped on thecurrent serving cell for more than one second.





Neighbour Cell Measurements(Based on HSC measurement rules)

UE applies cell selection criterion ‘S’Squal > 0 and Srxlev > 0 where:



Cells tested against the ‘H’ criterion to determine if HCS prioritized ranking should apply

Hs = Qmeas,s + Qhcss

Hn = Qmeas,n + Qhcsn –(TO n x Ln)

Cells meeting the ‘S’ and ‘H’ criterion and with the highest HCSpriority level are ranked using the ranking criterion ‘R’


Rn = Qmeas,n + Qoffsets,n –(TO n x (1 –L n ))


it is ranked higher than the serving cell for a time greater than Treselection

the UE has been camped on the current serving cell for at least one second

Figure 17

Cell Reselection Criteria with HCS

SC2804/S5/v1.1





Key parameters for optimizing the cell reselection process for a single scenario are

those used to calculate the HCS priority criterion ‘H’ and the ranking criterion ‘R’.These parameters are as follows:

• Hs –calculated HCS priority applicability for the serving cell

• Hn –calculated HCS priority applicability for a neighbour cell

• Rs –calculated ranking value for the serving cell

• Rn –calculated ranking value for a neighbour cell

• Qmeas,s –can be set as either Qqualmeas or Qrxlevmeas (serving cell)

• Qmeas,n –can be set as either Qqualmeas or Qrxlevmeas (neighbour cell)• Qhcss –HCS priority applicability threshold for the serving cell

• Qhcsn –HCS priority applicability threshold for a neighbour cell

• TOn –Temorary Offset, 3, 6, 9, 12, 15, 18, 21 or infinite

• Qhysts –hysteresis value, 0 to 40 dB in steps of 2

• Qoffsets,n –offset value, –50 to 50

• Treselection –timer value, 0 to 31 seconds

Note that a single values of Qhcss and Qhysts are set for the serving cell, but thevalues of Qhcsn, TOn and Qoffsets,n can be set independently for each listedneighbour cell.





Neighbour Cell Measurements(Based on HSC measurement rules)

UE applies cell selection criterion ‘S’Squal > 0 and Srxlev > 0 where:



Cells tested against the ‘H’ criterion to determine if HCS prioritized ranking should apply

Hs = Qmeas,s + Qhcss

Hn = Qmeas,n + Qhcsn –(TO n x Ln)

Cells meeting the ‘S’ and ‘H’ criterion and with the highest HCSpriority level are ranked using the ranking criterion ‘R’


Rn = Qmeas,n + Qoffsets,n –(TO n x (1 –L n ))


it is ranked higher than the serving cell for a time greater than Treselection

the UE has been camped on the current serving cell for at least one second


Cell Reselection Criteria with HCS

SC2804/S5/v1.1







Neighbour Cell

(GSM)

Serving Cell(UMTS FDD)

SsearchRATm –used to control neig hbour cell measurements

SHCS,RATm –used to control neig hbour cell measurements

Slimit,SearchRATm –used to control neig hbour cell measurements

Qrxlevmin and Pcompensation –used to control neighbour cellmeasurements and consideration for ranking

Qhcsn –used to control applicati on of HCS rules for reselection

TOn –temporary offset to stop short term reselection

Qoffsets,n –used to control ranking of the considered neighbour cell

Measurements:Qqualmeas

Qrxlevmeas

Measurement:RSSI

UE uses RSSI

for Qrxlevmeas

and Qmeas,n

HCS PriorityLevel n

HCS PriorityLevel n, n –1

or n+1

Figure 18

Reselection to GSM with HCS

SC2804/S5/v1.1





Figure 19 is a map showing an area in a major city. The large road running

diagonally from the top to the bottom of the map is a main route through the city.There are two lanes of traffic in each direction. It is always busy, but in normalconditions the traffic is fairly free running with an average speed of 40 kph.

To the right of the main road is a popular shopping area. The main shops are locatedon the ground floors of buildings A, B, C and D. A number of these front onto area E,which is a square. Part of this contains grass and trees, the rest is an open pavedarea often used by street performers. Building A is a large hotel used by tourists andby business travellers.

To the left of the main road, buildings G, H and F are offices used by several largecompanies.

The area in the map and the surrounding areas are currently served by a three-sector UMTS macro cell located on the roof of building A. The effective radius of thismacro cell was planned to be approximately 2 km. When first implemented themacro cell was intended to operate with a maximum load factor of 65%. At first, thisprovided adequate capacity, but lately the number of UMTS subscribers hassignificantly increased for this operator. The load factor limit is frequently reached,resulting in a large amount of blocked traffic.

The operator intends to build two micro cells in the map area to carry localized trafficand reduce load on the macro cell. The intention is to build one micro cell in theshopping area to serve the square and the hotel. The second micro cell is to be builtin the office area to pick up traffic from building G,H and F.

Task

1 Your group should suggest appropriate positions for the micro cells. A power-law propagation model and load factor of 75% suggests a radius for the microcells of about 100 m with good building penetration.

2 The operator would like to utilize HCS prioritization for these micro cells. Your group should identify the key parameters that will need to be introduced for HCS to function effectively. For each parameter identify the considerations for selecting a value and, if possible, suggest a value that could be used. Theoperator is keen that UEs in vehicles on the main road should continue to behandled by the Macro cell and you should take this into account.

3 Your group should also consider if it would be possible to introduce the microcells without utilizing HCS parameters. What would be the advantages anddisadvantages of adopting this strategy?

5 EXERCISE 2 – CELL RESELECTION SCENARIOS





Shops

DC

B

AHotel

EOpen

Square

G H

F

Scale0 50 m 100 m

Figure 19

Exercise 2 – Cell Reselection Scenarios

SC2804/S5/v1.1





6.1 Introduction

RRC provides control of most radio access functions. A key aspect of this is accesscontrol. This enables the UMTS user to connect to the UMTS network in order to useUMTS services. Access control can be broken down into two main parts: admissioncontrol and congestion control.

6.1.1 Admission Control

The admission control function is located at the Controlling RNC (CRNC) or ServingRNC (SRNC). This depends on the admission function being performed. CDMAnetworks operate on a soft capacity concept; this means that new calls increase theinterference level for all other calls. This affects the quality of all calls. Admissioncontrol provides the ability to admit or deny new users, new RABs, or new radiolinks. The decisions are based on QoS requirements, interference, current loadconditions and resource measurements.

6.1.2 Congestion Control

Congestion control will monitor, detect and control situations when overload

conditions occur. Congestion occurs when the network has run out of or is runningout of resources. The function of congestion control is to bring the system back to astable state (as quickly as possible).

6 RADIO RESOURCE CONTROL (RRC) FUNCTIONS





RNC

CoreNetwork

Node B

Node B

Broadcast systeminformation

Node B

Uplinkinterference

Node B

Node B

Newusers

Downlinkpower

Handover

resources(radio links)

Radio AccessBearers

QoS

Figure 20

Key RRC Functions

SC2804/S5/v1.1





6.2 Cell Access Restrictions

One strategy for managing traffic load on a cell is to use access class restrictions.This is not intended for general use but rather for exceptional or possibly emergencyconditions.

A blanket restriction may be placed on a cell by indicating in system information thatit is barred. If this is done a parameter called T barred is included. This secondparameter specifies the minimum time (in seconds) that a UE must wait beforerechecking if the barring on a cell has been lifted.

More subtle restrictions can be applied by barring individual or groups of accessclasses. A UE’s access class is stored on the SIM card. All subscribers are randomlyallocated an access class in the range 0 to 9. The aim is that 10% of the subscriber population will belong to each access class. Some particular types of subscriber mayalso be allocated one or more of the special access classes from 11 to 15. This isalso stored as a parameter on the SIM card. These classes are intended for use bygroups as indicated in Figure 21. This second special access class is only applicablewhen a subscriber is in their home network.

Although no subscribers are allocated access class 10, it may be used by theoperator to restrict access for emergency calls. If access class 10 is barred then

subscribers with access class 0 to 9 and any subscribers without an IMSI (i.e. noSIM card) may not make emergency calls. Subscribers in access classes 11 to 15can still make emergency call unless their access class is also specifically barred.





Access classes 0 to 9 10

Randomly allocatedto all users

May be individually barred

Used to restrictemergency calls

May be individually barred

11 12 13 14 15

Emergency calls

For PLMN use

Security services

Public utilities

Emergency services

PLMN staff

Figure 21

Access Classes

SC2804/S5/v1.1





6.2.1 Considerations for the use of Reserved Cells

Typically, reserved cells may be used to test potential sites for new base stations or to test system performance in respect of the introduction of new services. Additionally, existing cells may be temporarily put into this state for operationalreasons.

There are two possibilities for cell reservation. System information may indicate thatthe cell is ‘reserved for operator use’ or it may indicate that the cell is ‘ reserved for future extension’. In the case of reserved for operator use, UEs of access classes 11or 15 and those for whom this is the HPLMN will treat the cell as normal for allprocesses relating to cell selection and reselection. UEs with lower access classesand access classes 12 to 14 will treat the cell as barred. In the case of reserved for future extension, all UEs will treat the cell as barred.

This function is useful for testing purposes and for the integration of new cells.However, caution must be exercised with its use because it may limit normal serviceprovision in the vicinity of the reserved cell. Consider the UE shown in Figure 22.The cell in the example is marked as reserved for operator use. The UE has accessclass 9 so it cannot access the reserved cell. If the UE was to select another bordering cell on the same frequency as its serving cell there is a possibility that itcould cause excessive interference to the reserved cell. To prevent this the intra-

frequency cell reselection indicator may be set to ‘not allowed’. The UE will be forcedto search for a neighbour cell on another frequency.

There are two potential problems with this. Firstly another frequency implies another HCS layer, which may be inappropriate for the UE’s position or mobility state.Secondly, if the UE cannot find a suitable cell on another frequency it will camp on anacceptable cell and assume the limited service state. Thus service could be deniedto the user until the UE moves to another position even though there is not aproblem with coverage in the area.





UE accessclass 9

Reserved CellReserved for operator use

Serving Cell

Reserved Cells may be:Reserved for operator useReserved for future extension

Possibleinterference

Figure 22

Use of Reserved Cells

SC2804/S5/v1.1





6.3 Admission Control

The aim of admission control is to meet the QoS requirements of services providedwith new channel requests and at the same time to control the load in the UTRAN.Each new connection contributes to the noise rise in the serving cell and inneighbour cells. Admission control will need to estimate how much interferenceincrease there will be for a new request and how this will impact other connectionsalready established in the UTRAN.

There are many aspects to this and the algorithms used are not defined in the UMTSstandards. However, key parameters are:

• service mapping to channel types

• service mapping to QoS classes

• total downlink transmit power

• maximumm channel code power

• maximum allowable uplink transmit power

• maximum load factor

• current load

• UE capability

• code availability (OVSF code tree)

• channel element availability

Each of these factors, and perhaps others, will need to be included in the admissioncontrol process when it is used to determine if a newly-requested connection can besupported. Some or all of these factors may be available to the optimizer asadjustable parameters, or as new features that could be implemented. However, careshould be taken when changing any of these parameters since they are likely tohave been an input to the planning process. Significant changes in these values will

need extensive simulation to assess any likely wanted or unwanted effects.





Connection

request

service mapping to channel typesservice mapping to QoS classes

total downlink transmit power

maximum channel code power

maximum allowable uplink transmit power

maximum load factor

current load

UE capability

code availability (OVSF code tree)

channel element availability

Figure 23

Admission Control

SC2804/S5/v1.1





6.3.1 Call Admission Control (CAC) Prioritization

The way in which different traffic types are prioritized for consideration by admissioncontrol is an implementation issue. It may be subject to optimization, but this will be acollaborative process involving many aspects of system and service package design.However, once the requirements are set it will be a key aim of the overall radionetwork optimization process to see that these requirements are met for as much of the time as possible.

3GPP 25.9221 provides examples of how service classes may be utilized. One of these is shown in Figure 24. Allocation of resources would be prioritized according tothe QoS requirements associated with the requested service. Typically it is assumedthat connections for circuit-switched services will have priority over those for packet-switched services. In respect of packet-switched connections, those needing real-time delay constraints will have priority over those requesting non-real time.

It is important to ensure that services are matched appropriately to available QoSclasses. If this is not considered resource utilization may be inefficient.

For example, it may be desirable from a user-experience point of view to provide aweb browsing facility based on a guaranteed bit rate utilizing the conversational QoSclass. In the early operational stages of a UMTS network this should not cause a

problem since there will be ample capacity on most cells. Yet, as network loadincreases, new connection requests for the interactive class may be refused becausea guaranteed bit rate cannot be provided within the limits set for load factor.

As il lustrated in Figure 24, if the web browsing service was mapped to thebackground or perhaps the interactive QoS class it may not be refused even at hightraffic loads. This is because these classes will use spare capacity on the cell andresources are allocated dynamically without guaranteed bit rate or delay. For aservice such as web browsing this would be acceptable.

1 3GPP TS 25.922 Radio Resource Management Strategies.





Example Service Type Mapping to QoS Class (3GPP 25.922)

Example Service Mapping to Service Type and CAC Strategy (3GPP 25.922)

Type of Service QoS Class Delay Guaranteed bit rate

Premium Conversational Low Yes

Assured service Streaming Medium Yes

Best effort Interactive/background – No

Service CN Domain Type of Service CAC performed

CS Premium YesVoice

PS Premium Yes

PS Assured service YesWeb

PS Best effort No

Loadfactor

Time

Target load factor

Available forconnectionsneeding besteffort

Used byconnectionsneeding real-time guaranteed

bit rates

Figure 24

Call Admission Control (CAC)

SC2804/S5/v1.1









SECTION 6

CONNECTED MODE AND RADIO

LINK CONTROL

i









1 RRC CONNECTED MODE STATES 6.11.1 States for Circuit-Switched Operation 6.11.2 States for Packet-Switched Operation 6.11.3 UE Activity in Connected Mode 6.3

2 Channel Types 6.7

2.1 Channel Allocation and Dimensioning 6.92.2 RACH for Data Transfer 6.112.3 CPCH for Data Transfer 6.112.4 FACH for Data Transfer 6.112.5 DSCH for Data Transfer 6.132.6 DCH for Data Transfer 6.13

3 Power Control 6.153.1 Measurements for Power Control 6.153.2 Power Control for CPCH 6.173.3 Closed Loop Power Control 6.19

3.4 Outer Loop Power Control 6.253.5 Power Offsets 6.273.6 Transmit Power Control Headroom 6.293.7 Power Control in Soft Handover 6.313.8 Processing TPC Bits from a Single Physical Channel 6.333.9 Processing TPC Bits From Multiple Physical Channels 6.35

4 Exercise 1 – Power Control Scenarios 6.37

5 Soft Handovers 6.395.1 Measurements for Handover 6.395.2 Neighbour Cells for Soft Handover 6.415.3 Considerations for Active Set Size 6.435.4 Configuring the Measurement Message 6.455.5 Soft Handover Parameters and Triggers 6.475.6 Optimizing Soft Handover Regions 6.555.7 Inter-Cell Synchronization 6.71

6 Hard Handovers 6.736.1 Compressed Mode Measurements 6.736.2 Neighbour Cells for Hard Handover 6.79

6.3 Hard Handover Parameters and Triggers 6.81

SECTION CONTENTS

iii










• list and describe the RRC states within RRC connected mode

• suggest appropriate channels for user traffic transfer with different QoS

requirements

• describe User Equipment (UE) activity while in connected mode, both for

packet data transfer and circuit-switched data transfer

• summarize the key characteristics, configurations, capabilities and capacity

of each traffic-carrying channel available on a UMTS air interface

• describe the measurement process for assessing the requirement for power

control and handover commands

• describe the operation and controlling parameters for closed loop power

control

• characterize the effect of each parameter relating to closed loop power

control

• describe the operation of soft and hard handovers in UMTS

• characterize the effect of each parameter relating to hard and soft handover

control

• describe the operation of and options for compressed mode

• describe how the soft and hard handover processes may relate to the use of

hierarchical cell layers

• describe the interactions between UMTS and GSM/GPRS in respect of hard

handover

SECTION OBJECTIVES

v





The UE will enter the RRC Connected state by executing the RRC Connection

Procedure. This will place the UE in a connected mode state. One connected-modestate may be used for circuit-switched operation, i.e. the CELL_DCH state, whilethere are four possible connected mode states for packet-switched operation:CELL_DCH, CELL_FACH, CELL_PCH and URA_PCH.

1.1 States for Circuit-Switched Operation

For circuit-switched operation the UE will be placed in the CELL_DCH state. Here,the UE will be using DCCH or DTCH logical channels mapped to dedicated transportchannels carried by dedicated physical channels.

The UE will be known to the UTRAN at cell level according to its current active set.

1.2 States for Packet-Switched Operation

For packet-switched operation the UE may be placed in the CELL_FACH state. Nodedicated channels will be assigned to the UE, but common transport channels suchas RACH/FACH and CPCH/FACH may be used.

The UE will be known to the UTRAN at cell level and will perform cell updates.

The UE may be placed into the CELL_PCH state to avoid the need to constantlymonitor the downlink FACH channel. This will allow the UE to use DiscontinuousReception (DRX), prolonging battery life. The only way the UTRAN can reach the UEis by paging it at a cell level. Therefore the UE must still perform cell updatesinvolving the transition to the CELL_FACH state.

To minimize the number of cell updates the UE may be placed into the URA_PCHstate. Within a UTRAN Registration Area (URA) the UE may perform cell reselectionwithout performing a cell update unless a cell belongs to another URA. This will

invoke the URA Update procedure carried out in CELL_FACH state. To reach a UEthe UTRAN will page across the URA.

For large volumes of packet data the UE may be placed in the CELL_DCH state. TheUE will be known to the UTRAN at a cell level, but the UTRAN will control which cellsare to be used based upon the measurement information supplied by the UE.

1 RRC CONNECTED MODE STATES





CELL_DCH

CELL_PCH URA_PCH

CELL_FACH

Figure 1

RRC Connected Mode States

SC2804/S6/v1.1





1.3 UE Activity in Connected Mode

1.3.1 CELL_DCH

In the CELL_DCH state dedicated resources are allocated to the UE for circuit-switched data or packet-switched data. For packet-switched services the currentuplink and downlink data rates are indicated by the Transport Format CombinationIndicator (TFCI). If the allocated resources are insufficient to match the QoSrequirement the network will initiate a reconfiguration of the transport format.

In the uplink direction the UE can report the observed traffic volume to the network inorder for the network to re-evaluate the current allocation of resources. In this waythe UE connected state may change from CELL_DCH to CELL_FACH or vice versa.

In the CELL_DCH state the UE performs measurements and transmits measurementreports based upon the measurement control information.

Certain FDD UEs can read system information in the CELL_DCH state using FACH.

1.3.2 CELL_FACH

In this state no dedicated resources are allocated to the UE. Instead the UE monitorsthe downlink FACH continuously. The UE may be assigned a common transportchannel, e.g. RACH or CPCH, which can be used at any time.

Before data is transmitted in the uplink direction the UE reports the observed trafficvolume to the network in order for the network to re-evaluate the current allocation of resources. A selection procedure then determines whether the information should besent on a common transport channel or whether a transition to CELL_DCH isrequired.

In the CELL_FACH state the position of the UE is known at cell level. The cell update

procedure must be executed if the UE reselects a new cell. Data transmission in thedownlink direction can begin without prior paging.

The UE will monitor system information broadcasts on BCCH. The measurementcontrol information broadcast on BCCH informs the UE about measurements andreporting.





dedicated resources for PS and CS dataresources allocated to match QoS requirements

UE reports observed traffic volume

UE performs measurements and transmits reports

certain UEs read system information on FACH

no dedicated resources

UE constantly monitors FACH

UE reports observed traffic volume before data transmission

UE known at cell level and performs cell updates

UE reads system information on BCH

CELL_DCH

CELL_FACH

Figure 2

CELL_DCH and CELL_FACH States

SC2804/S6/v1.1





1.3.3 CELL_PCH

In the CELL_PCH state no resources are allocated to the UE. The UE may performDRX and must be paged in the cell to initiate downlink data transfer. Uplink activitywill trigger a move to the CELL_FACH state.

The UE will be known to the UTRAN on a cell level, therefore the UE must performthe cell update procedure when reselecting a cell. This will be done in theCELL_FACH state. To reduce the number of cell updates the UE may be ordered tothe URA_PCH state. This will be done while in the CELL_FACH state. It can bebased on an inactivity timer, and optionally a counter. The UTRAN can order the UEto the URA_PCH state when the number of cell updates exceeds a certain threshold.

The UE performs measurements and transmits measurement reports according tothe measurement control information.

The UE will also read system information broadcasts on the BCH.

1.3.4 URA_PCH

In the URA_PCH state no resources are allocated to the UE. For data transmission a

transition to the CELL_FACH state is required. The UE may use DRX and must bepaged across the URA to initiate downlink data transfer. Uplink activity will trigger amove to the CELL_FACH state.

The UE will be known to the UTRAN at the URA level. If the UE selects a cellbelonging to another URA it must perform the URA update procedure in theCELL_FACH state.

The UE performs measurements and transmits measurement reports according tothe measurement control information.





CELL_PCH

no dedicated resources allocatedUE may implement DRX

UE performs cell updates

UE paged at cell level

UE moves to CELL_FACH for data transfer


URA_PCH

no dedicated resources allocated

UE may implement DRX

UE paged over URA

UE performs fewer cell updates

UE moves to CELL_FACH for data transfer


Figure 3

URA_PCH and CELL_PCH States

SC2804/S6/v1.1





There are a variety of channel types available for data transfer. The choice largely

depends upon the QoS requirements. In the UMTS QoS concept four traffic serviceclasses are defined:

• conversational

• streaming

• interactive

• background

The different classes indicate the sensitivity to delay where conversational class is

the most delay sensitive and background class is the least delay sensitive.Conversational and streaming classes are intended to carry Real-Time (RT) servicesand may be circuit switched or packet switched. Interactive and background classesare intended to carry Non-Real-Time (NRT) services over a packet-switchedconnection.

The data channels for the UMTS air interface can be divided into Dedicated,Common and Shared. Dedicated Channels (DCH) resemble circuit-switchedconnections and are therefore most suitable for real-time services. CommonChannels (RACH/FACH/CPCH) are contention based and are therefore mostsuitable for the bursty packet data found with some non-real-time services. Theshared channels (DSCH) have similar properties to both common and dedicatedchannels but are likely to be used for non-real-time services.

2 CHANNEL TYPES





SensitivityTo Delay

ConversationalClass

StreamingClass

InteractiveClass

BackgroundClass

Real-TimeServices

CS/PS

Non-Real-TimeServices

PS

DedicatedChannels

DCH

CommonChannels

RACHFACHCPCH

SharedChannels

DSCH

Traffic ServiceClass

ServiceType Transport Channels

Figure 4

Channel Types

SC2804/S6/v1.1





2.1 Channel Allocation and Dimensioning

The DSCH and the CPCH are optional channels leaving the RACH/FACH pair for non-real-time services. The RACH transport channel is carried uplink over the air interface in the PRACH. There are only 16 PRACH scrambling codes available per cell, which limits the number of RACH/FACH pairs that could be used for non-real-time services.

The DCH may support real-time and non-real-time services. There are a largenumber of UE scrambling codes; the only limitation is the number of availablespreading codes. There are few low spreading factor codes and a larger number of high spreading factor codes. The allocation of codes is the responsibility of the RNC.

The selection of channel type is done by the RNC and is based upon:

• QoS

• data volume

• traffic load

• interference level

• performance of the transport channel





DSCH + CPCHoptional

RACH/FACHnon real-time

services

DCHreal-timeservices

16 RACH/FACHpairs

Large number of codes

Leaving

Choice performed by RNCbased upon

QoS

data volume

traffic load

interference level

performance of transport channel

?

Figure 5

Channel Allocation and Dimensioning

SC2804/S6/v1.1





2.2 RACH for Data Transfer

The RACH is an uplink common transport channel paired with the FACH in thedownlink direction. This would be used in the CELL_FACH state to transfer smallvolumes of packet data. The duration of transmission can be set to 10 ms or 20 msdepending upon cell size. For large cells the 20 ms setting corresponds to a bit rateof 8 kbit/s giving a good processing gain as is required for mobiles at the cell edge.However, open loop power control is used on RACH and a 20 ms burst will generatemore noise in a cell compared to 10 ms.

As well as the absence of fast power control, soft handover is not supported. Thereare also a limited number of fixed spreading factor codes for the RACH. However,the RACH offers short set-up times.

2.3 CPCH for Data Transfer

The CPCH is an uplink common channel paired with FACH in the downlink direction.This would be used in the CELL_FACH state to transfer small to medium volumes of packet data. The duration of transmission is controlled by the RNC and a number of fixed spreading codes are available in the cell. There is a limit on the number of Physical Common Packet Channel Channel scrambling codes of up to 64 in a cell.

The CPCH supports fast power control and offers fast set up time, but does notsupport soft handover. It therefore only offers reasonable radio performance.

2.4 FACH for Data Transfer

The FACH is a downlink common channel that is usually paired with RACH or CPCH. It offers a fixed number of channelization codes and is best suited to carryingsmall volumes of bursty data. The absence of fast power control and soft handover means its radio performance is low.





Channel Type

UE State

Direction

Power Control

SHO

Data Volume

Set-up Time

RadioPerformance

RACH

Common

CELL_FACH

UL

Open loop

No

Small bursty

Low

Poor

CPCH

Common

CELL_FACH

UL

Closed loop

No

Small/mediumbursty

Low

Medium

FACH

Common

CELL_FACH

DL

Open loop

No

Small bursty

Low

Poor

DSCH

Shared

CELL_DCH

DL

Closed loop

No

Medium/highbursty

High

Medium

DCH

Dedicated

CELL_DCH

DL/UL

Closed loop

Yes

Medium/highprolonged

High

Good

Figure 6

Channels for Data Transfer

SC2804/S6/v1.1





2.5 DSCH for Data Transfer

The DSCH would be used in the CELL_DCH state. Channelization codes would beshared between a number of users allowing a larger number of users to access high-bit-rate services. The DSCH is paired with a DCH downlink offering fast power control with the UEs using a DCH uplink. The DSCH is best suited to medium to highdata volumes that are bursty in nature, but it demands a longer set-up time. Theabsence of soft handover gives a medium radio performance.

2.6 DCH for Data Transfer

The DCH is used in the CELL_DCH state. The DCH is used in both uplink anddownlink directions and supports a wide variety of bit rates. It supports fast power control and soft handover, and is ideal for carrying medium to large volumes of data.The DCH offers good radio performance but it is not suited to bursty data.





Channel Type

UE State

Direction

Power Control

SHO

Data Volume

Set-up Time

RadioPerformance

RACH

Common

CELL_FACH

UL

Open loop

No

Small bursty

Low

Poor

CPCH

Common

CELL_FACH

UL

Closed loop

No

Small/mediumbursty

Low

Medium

FACH

Common

CELL_FACH

DL

Open loop

No

Small bursty

Low

Poor

DSCH

Shared

CELL_DCH

DL

Closed loop

No

Medium/highbursty

High

Medium

DCH

Dedicated

CELL_DCH

DL/UL

Closed loop

Yes

Medium/highprolonged

High

Good

Figure 6 (repeated)

Channels for Data Transfer

SC2804/S6/v1.1





In a CDMA system, where all the users share a common frequency, interference

control is of paramount importance. From an uplink point of view the near–far effectmust be controlled to avoid weaker, distant users being drowned out by nearer,stronger users. From a downlink point of view, system capacity gains can beachieved by ensuring that downlink channels use the minimum required power.

Power control in UMTS is divided into open loop power control and closed looppower control. The open loop mechanism is used to set the initial transmit powers oncell access. Closed loop power control (often called fast power control) dynamicallychanges the transmit power levels 1500 times a second. Closed loop power controlconsists of inner loop and outer loop processes.

3.1 Measurements for Power Control

There are two measurements required for power control. Firstly, the UE mustmeasure and report the RSCP. measured on the CPICH. This is defined as thereceived power on one code measured on the PCPICH, the reference point beingthe antenna connector of the UE.

Secondly, the Node B must measure the received total wideband power, which is thewideband power including noise generated in the receiver. The reference point is the

receiver antenna connector.

3 POWER CONTROL





Closed Loop

Outer Loop Inner Loop

SetsSIRTargets

Perform fastpower control

Open Loop

Sets initialtransmit

power levels

UMTS Power Control

Figure 7

Power Control

SC2804/S6/v1.1





3.2 Power Control for CPCH

The CPCH is an uplink contention-based channel that uses open loop power controlinitially before the fast closed loop process takes over.

The first part of the CPCH access procedure is identical to the RACH procedure. TheUE first determines the Preamble_Initial_Power which will set the initial open looppower level for the first CPCH access preamble (PRACH = PCPCH).

Successive access preambles will be transmitted at an increased power leveldetermined by the power ramping factor Power_Ramp_Step, ∆P0, an integer value >0.

Once the UE has received a positive acknowledgement on the AICH, a collisiondetect preamble is transmitted at the same power level as the successful accesspreamble. If this is not positively acknowledged the CD-preamble power is rampedup by ∆P1 and this new transmit power level is used in place of Power_Ramp_Step(∆P0) for a time period.

On receipt of the Postive acknowledgement on the CD/CA-ICH channel the CPCHprocedure may enter the power control preamble phase or begin transmitting theCPCH. In either case the transmit power level is increased by a power offset

measured in dB. (Pmessage-control – Pcd).

The purpose of the power control preamble is to rapidly adjust the transmit power level to the optimal setting using the fast power control algorithms with different stepsizes. Only the control part of the PCPCH is affected. The normal step sizes will beused after the first eight slots or if the power control command reverses for the firsttime. Then both control and data parts will be power controlled by the normal closedloop process.

After the first slot in the power control preamble, changes in the control part of PCPCH will be determined by:

∆PCPCH-CP = ∆TPC-init x TPC_cmd

Using power control algorithm 1, ∆TPC-init is equal to the minimum value out of 3 dBand 2∆TPC.

Using power control algorithm 2, ∆TPC-init is equal to 2dB.

TPC-cmd is derived according to algorithm 1 irrespective of which algorithm is used.





Power

∆P0PRACH

∆P0

∆P1 ∆Pp-m

Data Part

Control Part

Time Access preamble

CD-preamble CD-preamblein case of no ACK

Power control

preamble

0 or 8slots

PRACH = PCPCH

∆P0

∆P1

∆PP-M

=

=

=

=

Initial open loop power level for first CPCH access preamble.

Power step size for each successive CPCH access preamble.

Power step size for each successive CPCH access preamble in case ofno AICH response. Only valid for a time period before being replaced by

∆P0.

Pmessage-control – Pcd, measured in dB. Power offset between CD-preamble

and the initial transmit power of the CPCH power control preamble (orthe control part if no power control preamble).

Figure 8

CPCH Power Control

SC2804/S6/v1.1





3.3 Closed Loop Power Control

Closed loop power control operates in both uplink and downlink directions.

3.3.1 Uplink Power Control

The initial uplink DPCCH transmit power is set by higher layers based upon thefollowing expression:

DPCCH_Initial_power = DPCCH_Power_offset – CPICH_RSCP

Where CPICH_RSCP is measured and reported by the UE andDPCCH_Power_offset is calculated in the RNC from the following:

DPCCH_Power_offset = CPICH_TX_power + UL interference + SIRDPCCH –10log(SFDPCCH)

Where the UL interference is the received total wideband power. SIRDPCCH is the SIRvalue determined by the RNC and SFDPCCH is the spreading factor for DPCCH.

Because the spreading factor for DPCCH and DPDCH is not necessarily the same,

gain factors are applied to each channel to balance the power allocated to eachchannel.

βc is the gain factor for DPCCH and βd is the gain factor for DPDCH

The gain factors can either be signalled to the UE from higher layers for a certaintransport format combination or calculated by the UE.





DPCCH_Initial_power = DPCCH_Power_offset - CPICH_RSCP

where CPICH_RSCP is measured and reported by the UEand

DPCCH_Power offset = CPICH_TX_power + UL interference + SIRDPCCH - 10log (SFDPCCH)

where UL interferenceSIRDPCCH

SFDPCCH

= total wideband power = SIR determined by RNC= Spreading factor

Figure 9

UL Closed Loop Power Control

SC2804/S6/v1.1





3.3.2 Uplink Power Adjustment

The uplink inner loop power control adjusts the UEs transmit power to keep thereceived uplink SIR equal to the SIR target, SIRtarget. The SIRtarget is set by the RNCusing the outerloop power control process and signalled to the Node B.

Higher layer signalling will inform the UE of which power control algorithm should beused. Additionally higher layer signalling will indicate the TPC-Step Size which isused to set TPC. If the TPC-StepSize is ‘dB1’ then TPC is set to 1 dB. If the TPC-StepSize is set to ‘dB2’, then TPC is set to 2 dB.

After determining the TPC-cmd from the TPC bits transmitted downlink, the UE willalter its transmit power according to the following:

∆DPCCU = ∆TPC x TPC – cmd





x TPC_cmd

where TPC_cmd = +1, –1, 0

Figure 10

UL Closed Loop Power Control

SC2804/S6/v1.1






3.3.3 Downlink Power Control

The downlink inner loop power control adjusts the Node B’s transmit power in order to keep the downlink SIR equal to the SIRtarget. The outer loop process sets theSIRtarget which is given to the UE by higher-layer signalling.

The UE estimates the received downlink power on the DPCH and estimates the levelof interference in the cell. The UE can then determine the SIRest.

If SIRest > SIRtarget set the TPC command to ‘0’ indicating power down.

If the SIRest

< SIRtarget

set the TPC command to ‘1’ indicating power up.

Within the Node B the TPC commands are interpreted depending upon theparameter DPC_MODE. If DPC_MODE is set to 0 the TPC commands from the UEwill be estimated TPCest to be 0 or 1 and will change the power in every slot. If theDPC_MODE is set to 1 the Node B will estimate the commands over three slots tobe 0 or 1 and will change power every three slots.

The power control step size ∆TPC can be set to 0.5, 1, 1.5 or 2 dB. A value of 1 dBis mandatory.




SIRest

SIRtarget

Figure 11

DL Closed Loop Power Control

SC2804/S6/v1.1





3.4 Outer Loop Power Control

The uplink and downlink outer loop power control is executed by vendor-specificalgorithms. The uplink procedure is executed in the SRNC and is responsible for setting the SIRtarget in the Node B for each individual closed loop power controlprocess. The downlink procedure is performed in the UE for each transport channel.

The SIRtarget changes as the UE speed and multipath propagation environmentchange. The greater the variation in received power level the greater the SIRtarget

needs to be. The target is determined according to the estimated link quality, basedupon BER or Block Error Rate (BLER). A Cyclic Redundancy Checksum (CRC)could be used to determine if the target should be increased or decreased, e.g. if theCRC is OK the target can be lowered, otherwise it is increased. Suggested values for the step size range from 0.1 to 1.0 dB





TPCTransmits Power Control Commands

Outer Loop

Inner Loop

SIRtarget

Measurementreport required

Measurement reportslayer 3 information

Outer loop power control sets Signal-to-Interference Ratio (SIRtarget)

Inner loop power control in Layer 1 adjusts peer entity transmit power so that themeasured SIR fulfills SIRtarget requirements

TPC

TPC

Figure 12

Closed Loop Power Control

SC2804/S6/v1.1





3.5 Power Offsets

When a UE is in soft handover it can be proven that allocating more power to theTPC bits in the downlink DPCCH relative to DPDCH can result in 0.4 to 0.6 dBreduction in UE transmit power.

The reason for this is that allocating proportionally more power to the TPC bitsimproves the power control signalling quality. There are three power offsets that canbe applied to the downlink DPCH: PO1 sets the offset between DPDCH and theTFCI bits, PO2 is the offset for the TPC bits, and PO3 sets the offset for the pilot bits.





Data 1

DPDCH DPCCH DPDCH DPCCH

PO3

PO1

PO2

Data 2

Pilot

TFCI

TPC

Figure 13

Power Offsets

SC2804/S6/v1.1





3.6 Transmit Power Control Headroom

When determining the maximum cell radius in a CDMA system the maximum uplinkpath loss is used. This allows the link budget to be set and a fast fade margin to beadded. Fast power control in UMTS is able to follow the fast fading envelope,particularly at low terminal speeds. However, at the cell edge the mobile will betransmitting maximum power, i.e. with no headroom. For all other positions in the cellthe mobile will be transmitting power levels lower than the maximum so there will bevarying degrees of headroom.

This simplistic view does not reveal one of the hidden benefits of fast power control.Simulations have shown that for a speech service with a BLER of 1% without fastpower control the required Eb/No for a pedestrian at 3 km/h is 11.3 dB. With fastpower control the required Eb/No is 5.5 dB. The difference of 5.8 dB is known as the‘fast power control gain’ or ‘power control headroom’. This can be translated into a3.6 dB reduction in transmitted power. This gain diminishes with increasing speedbecause of the inability to follow the fading profile.

A better definition of transmit power control headroom is:

TPC headroom = average required received Eb/No without fast power control – average required received Eb/No with fast power control





Max power =TX power

Max power

TX power

Max power

TX power

Headroom Headroom

Node B

Figure 14

Power Control Headroom

SC2804/S6/v1.1





3.7 Power Control in Soft Handover

When a UE is in soft handover TPC bits are sent uplink and read by all the Node Bsengaged in the soft handover. If there are signalling errors on the air interface, oneNode B may interpret the TPC bits differently to another Node B. This may result inone Node B powering up and another powering down. The resulting difference inpower levels is known as ‘power drifting’. This has the effect of degrading the softhandover gain.

One way of combating power drifting is for the Node Bs to average the transmissioncode power levels of the connections engaged in soft handover and pass them to theRNC. The time used for the averaging process is the measurement reporting period,typically set to 500 ms. From these measurements the RNC calculates a referencepower level which is sent to all the cells concerned. The Node Bs will then use this tocalculate a small power adjustment towards the reference value, thereby reducingthe power drifting.





Calculate power reference levels

sent to all Node Bsinvolved in soft

handover

Node Bpowers up

Node Bpowers down

Power drifting

TPCcommands

TPCcommands

Averagetransmission

code power

Averagetransmissioncode power

Figure 15

Power Drifting

SC2804/S6/v1.1





3.8 Processing TPC Bits from a Single Physical Channel

When the UE is not engaged in a soft handover it will be required to process onlyone set of TPC bits in each slot. The TPC bits must be processed in order to derive avalue for TPC_cmd. There are two algorithms defined for this purpose. The UTRANwill indicate in higher-layer signalling which of the two algorithms should be used.

3.8.1 Algorithm 1 for Single Channels

In this case, the TPC bits are interpreted as either 1 or 0. These values are thendirectly mapped to values for TPC_cmd of +1 and –1 respectively.

3.8.2 Algorithm 2 for Single Channels

In this case, the algorithm represents an amalgam of five consecutive slots. For thefirst four received slots the value of TPC_cmd is set at zero, irrespective of TPC bitvalues. On receipt of the fifth slot, the five consecutive slots are considered together.TPC_cmd will only be set as +1 or –1 if all five slots are 1 or 0 respectively, otherwiseTPC_cmd remains set at zero. Thus a power control command is implemented onlythree times in each frame.





0 1 2 43 65 1413

TPC TPC TPC TPC TPC TPC_cmd

1 1 1 1 1

0 0 0 0 0

X X X X X

TPC_cmd = +1

TPC_cmd = –1

TPC_cmd = 0

Figure 16

Algorithm 2 for Single Channels

SC2804/S6/v1.1





3.9 Processing TPC Bits From Multiple Physical Channels

When the UE is engaged in a soft handover it will be receiving TPC bits from morethan one Node B. These TPC bits will need to be combined in order to derive a valuefor TPC_cmd. The two defined algorithms are also used for this purpose.

3.9.1 Algorithm 1 for Multiple Channels

In each slot period, soft decision decoding is used to read the TPC bits from eachNode B in the active set. The value of TPC_cmd will be set to 1 only if all the Node BTPC bits are 1; otherwise, the value of TPC_cmd will be set to –1.

3.9.2 Algorithm 2 for Multiple Channels

This process is outlined in Figure 15. In each slot period the TPC bits are decodedfor each of the active set Node Bs. This is repeated for five consecutive slot periods.On reception of the fifth slot all five slots are considered such that for each Node B avalue of TPC_temp

iis determined. The value of TPC_temp

iwill be set at +1 or –1

only if the five consecutive slots are all 1 or all 0 respectively; otherwise, it will takethe value 0.

The second step is for the UE to combine the values for TPC_tempiinto one value

for TPC_cmd. This is done using the relationship shown in Figure 15. The wholeprocess is repeated for each group of five slots, i.e. three times in each frame.





TPC TPCTPC i

i = 1

TPC TPC TPC_temp 1

TPC_temp 2 TPC_cmd

From Node B1

From Node B2

From Node BN TPC_temp N

TPC_tempi > 0.5

TPC_tempi < –0.5

TPC_cmd is set to 1 if

1 1 1 1 1 – TPC_temp = +1

0 0 0 0 0 – TPC_temp = –1X X X X X – TPC_temp = 0

TPC_cmd is set to –1 if

x

x

Otherwise, TPC_cmd is set to 0

N = Number of cells

TPC TPCTPC i

i = 2

TPC TPC

TPC TPC

i = N

TPC TPC

1

i=1N

N

1N

Σ

i=1

N

Σ

.....

0 1 2 3 4 5 6 7 8 9 10 11 12 13 14

TPCN

Figure 17

Algorithm 2 for Multiple Channels

SC2804/S6/v1.1





Power Control Exercise

A class 4 UE establ ishes a connection in a cell. Based upon the fo llowingparameters determine the uplink transmit power after 60 seconds of operation. At theend of this time period there is an excess of power down commands equivalent to0.05% of the total number of commands.

Class 4 UE = 21 dBmCPICH Tx Power = 33 dBmTarget SIR = 9 dBUL Interference = –98 dBmSpreading Factor = 64CPICH RSCP = –91 dBmPower Step Size = 1 dB

4 EXERCISE 1 – POWER CONTROL SCENARIOS









In a CDMA system the base stations are likely to be on the same frequency. It is

therefore possible for a UE to set up communication links with a number of basestations simultaneously and be power controlled by all the base stations. This isknown as a Soft Handover (SHO) and facilitates the movement of a connectedmobile. There are two types of SHO defined for UMTS, a soft handover and a softer handover. The former is when the UE is communication with two or more Node Bsand signalling and traffic is sent to the RNC for combining. This combining process isknown as macrodiversity. A softer handover occurs when a UE is in soft handover between cells controlled by the same Node B. In this case the signalling and traffic iscombined in the Node B and is known as microdiversity. As far as the UE isconcerned there is no distinction between a soft or softer handover.

5.1 Measurements for Handover

The measurement process for UMTS is more flexible than for GSM. When a UE is inthe CELL_FACH, CELL_PCH or URA_PCH states it will use the informationbroadcast as system information for measurements. These measurements willlargely be for cell reselection purposes. The only measurement to be reported will betraffic volume measurements and these will be sent in the CELL_FACH state.

When a UE is in the CELL_DCH state it will be told precisely what to measure andwhen to report the measurement data using the Measurement Control message

delivered to the UE via DCCH signalling. The Measurement Control messageincludes:

• measurement identity number, a reference used by the UTRAN

• measurement command, used to start, modify or suspend measurements

• measurement objects – neighbour cell information

• measurement quantity – what to measure, e.g. RSCP or RSSI

• measurement reporting quantities – what quantities to report

• measurement reporting criteria, which allow for the setting of triggers• reporting mode – acknowledged or unacknowledged mode of RLC

• one of seven measurement types

5 SOFT HANDOVERS





measurement identity number

measurement command

measurement objects

measurement quantity

measurement reporting quantities

measurement reporting criteria

reporting mode

measurement type – intra-frequency

– inter-frequency

– inter-system

– traffic volume

– quality

– internal – location measurements

UTRAN

Measurement Control

Message

Figure 18

Measurements for Handover

SC2804/S6/v1.1





5.2 Neighbour Cells for Soft Handover

For each defined cell in the RNC there will be a neighbour list. Potentially threeneighbour lists may be defined:

• intra-frequency list of up to 32 cells on the same frequency as the server

• inter-frequency list of up to 32 cells on a different UMTS frequency

• inter-system list of up to 32 GSM frequencies

The neighbour lists are broadcast as part of system information, but a UE in

connected mode can receive details using dedicated signalling via the DCCH.

To identify a UMTS neighbour the list must include the following information:

• Global RNC Id (MCC + MNC and RNC Id)

• Cell Identifier (CI)

• Location Area Code (LAC)

• Routing Area Code (RAC)

• UARFCN

• Scrambling code for the PCPICH

For a GSM cell the following information will be required:

• Cell Global Identity (CGI)

• BCCH frequency

• Base Station Identity Code (BSIC)





Neighbour List

Intra-frequency

list of up to 32 cells

Inter-frequency


Inter-system


For each defined

UMTS cell

Gobal RNC Id

Cell Id

LAC

RAC

UARFCN

Scrambling code for PCPICH

CGI

BCCH frequency

BSIC

UMTSneighbour

list

GSMneighbour

list

Figure 19

Neighbour Cells for Soft Handover

SC2804/S6/v1.1





5.3 Considerations for Active Set Size

The cells that are engaged in soft handover are known as the active set of cells. TheUE will be told which neighbours to perform measurements on as well as triggers for reporting. If the measurements on a neighbour satisfy the reporting criteria, ameasurement result is sent to the RNC. If resources are available in the target cellthe active set update procedure will be activated. The new cell is added to the activeset and the UE is engaged in a soft hand over with the new cell. This is known as aMobile Evaluated Hand Over (MEHO). However, decisions about handovers are stillmade by the RNC.

There are a number of considerations for determining the size of the active set.Firstly, each cell in the active set will require a Radio Link (RL). RLs are added for each cell in the active set. The maximum number of RLs is eight. 3GPP 25.133specifies a minimum of six.

Secondly, a UE can only engage in soft handover with another base station if it has aspare finger in the rake receiver. The maximum number of fingers in a UE rakereceiver is not specified and is manufacturer dependent.

The soft handover probability target set in the radio network planning should be keptbelow 30–40% for the following reasons:

Mobiles engaged in soft handover will consume more downlink spreading codes thansingle link connections. Reserving spreading codes in a Node B for soft handover willimpact capacity.

Each RL that is established will also require resources on the Iub interfaces. A 40%probability of soft handover demands 40% extra backhaul capacity. For UEsengaged in softer handover there will be no impact on backhaul capacity becausesignalling and traffic will be combined locally in the Node B.

The benefit of soft handover is soft handover gain. A UE can combine a number of

downlink signals using the rake receiver and get a net improvement in performanceof as much as 3 or 4 dB compared to a single link connection. This can be taken intoaccount favourably when determining the link budget. However, a UE in softhandover will also be power controlled by all the Node Bs concerned. If the path lossis the same for all the Node Bs the soft handover gain will be optimal. But if there is asmall difference in path loss figures of a few decibels then the UE is likely to bepowered up diminishing the soft handover gain. The subsequent noise rise willdiminish capacity.

Simulations (3GPP 25.942) have shown that in a planned area only 1% of locations

require SHO to seven or more cells. Additionally, the SHO gain is minimal whenmore than three cells are in the active set. The conclusion is that the UE does nothave to support more than four to six cells in the active set.





Considerations

for Active Set Size

SoftHandover

Gain

Maximum No.

of Radio Links= eight

Number of Rake Fingers

SpreadingCode

Capacity

BackhaulCapacity

Figure 20

Considerations for Active Set Size

SC2804/S6/v1.1





5.4 Configuring the Measurement Message

The configuration of the measurement message depends on what was specified inthe downlink Measurement Control message. The reporting procedure is initiated bythe UE when the reporting criteria are met, which can be based upon triggers or timers.

The measurement message is used in both the CELL_DCH and the CELL_FACHstates. However, in the CELL_FACH state only traffic volume measurements arereported. UEs in the URA_PCH or the CELL_PCH state will transit to theCELL_FACH state before transferring traffic volume information.

The measurement report will contain the measured results of the quantity indicatedin the Measurement Control message. The list will be ordered from best cell to worstcell (where the best cell is the one with the highest Ec/No value or smallest pathloss).

The measured results include:

• intra-frequency measured results list

• inter-frequency measured results list

• inter-RAT measured results list

• traffic volume measured results list

• quality measured results list

• UE internal measured results

• UE positioning measured results

Details about these can be found in 3GPP 25.331, but as an example the intra-frequency measured results list contains:

• the scrambling code on the PCPICH

• CPICH Ec/No or CPICH RSCP or pathloss

• optionally, the cell ID, the SFN-SFN observed time difference and cellsynchronization information





Message Type

Integrity Check Info

Measurement Identity

Measured Results

Measured Resultson RACH

Additional MeasuredResults

> Measured Results

Event Results

OP

OP

MP

OP

Intra-frequencyMeasured Results

List

Inter-frequencyMeasured Results

List

Inter-RATMeasured Results

List

Traffic VolumeMeasured Results

List

Quality MeasuredResults List

UE InternalMeasured Results

List

UE PositioningMeasured Results

List


List

> Cell MeasuredResults

OP

MP

Measured Results

Cell Identity

SFN-SFNObserved Time

Difference

CellSynchronization

Information

PCPICH

Info

CPICH Ec/No

CPICH RSCP

Path loss

Cell MeasuredResults

OP

OP

OP

MP

OP

OP

OP


List

Measurement Report

AM or UM RLCDCCH

Figure 21

Configuring the Measurement Message

SC2804/S6/v1.1





5.5 Soft Handover Parameters and Triggers

In UMTS there are a number of triggers associated with the measurement process.The 3GPP specifications do not give actual parameters for the thresholds, butequipment manufacturers are likely to implement thresholds as parameters within theUTRAN. The triggers can be grouped into six categories that tie in with six classes of measurement reports, and include:

• intra-frequency measurements

• inter-frequency measurements

• inter-system measurements• traffic volume measurements

• quality measurements

• UE internal measurements

5.5.1 Intra-frequency Measurements

The measurement quantity that can be used to evaluate an intra-frequency eventincludes the Ec/No measured on the PCPICH, RSCP after despreading or the

downlink path loss calculated as:

Path loss in dB = PCPICH TX Power – PCPICH RSCP

Path loss values will be rounded up to the nearest integer value and range between46 and 158 dB.

Six different events may trigger an intra-frequency measurement report. However,reports can be sent periodically if no new cells have been added to the active set.The six events are:

• Event 1a: a PCPICH enters the reporting range

• Event 1b: a PCPICH leaves the reporting range

• Event 1c: a non-active PCPICH becomes better than an active PCPICH

• Event 1d: a change of best cell

• Event 1e: a PCPICH becomes better than the absolute threshold

• Event 1f: a PCPICH becomes worse than the absolute threshold





Intra-frequency measurementsInter-frequency measurementsInter-system measurementsTraffic volume measurementsQuality measurementsUE internal measurements

May be used to

trigger handover

Figure 22

Measurements to Trigger Handover

SC2804/S6/v1.1





5.5.2 Event 1a: A PCPICH Enters the Reporting Range

A report will be triggered if the following is fulfilled:

Triggering condition for path loss

Triggering condition for all the other measurement quantities

Where:

MNew is the measurement result of the cell entering the reporting range.

Mi is a measurement result of a cell in the active set.

N A is the number of cells in the current active set.

For path loss:

MBest

is the measurement result of the cell in the active set with the lowestmeasurement result.

For other measurements quantities:

MBest is the measurement result of the cell in the active set with the highestmeasurement result.

W is a parameter sent from the UTRAN to the UE.

R1a is the reporting range constant.

H1a

is the hysteresis parameter for the event 1a.

If the measurement results are path loss or CPICH Ec/No then MNew, Mi and MBest

are expressed as ratios.

If the measurement result is CPICH-RSCP then MNew, Mi and MBest are expressed inmilliwatts.


/2)H(RW)10LogM(1)(1/M1/W10Log10LogM1a1aBest

N

1i

iNew

A

−+−+

≤ ∑

=

/2)H(RW)10LogM(1MW10Log10LogM1a1aBest

N

1i

iNew

A

−−−+

≥ ∑

=




Time

AbsoluteThreshold

PCPICH

Ec/No

ReportingRange

CPICH 2

CPICH 1

Event

1a

Figure 23

Event 1a: PCPICH Enters Reporting Range

SC2804/S6/v1.1





5.5.3 Event 1b: A PCPICH Leaves the Reporting Range

A report will be triggered when the following equation is satisfied:

Leaving triggering condition for path loss

Leaving triggering condition for all the other measurement quantities

Where:

MOld is the measurement result of the cell leaving the reporting range.

Mi is a measurement result of a cell in the active set.

N A is the number of cells in the current active set.

For path loss:

MBest

is the measurement result of the cell in the active set with the lowestmeasurement result.

for other measurements quantities:

MBest is the measurement result of the cell in the active set with the highestmeasurement result.


R1b is the reporting range constant.

H1b

is the hysteresis parameter for the event 1b.

If the measurement results are path loss or CPICH Ec/No then MNew, Mi and MBest

are expressed as ratios.

If the measurement result is CPICH-RSCP then MNew, Mi and MBest are expressed inmilliwatt.


/2)H(RW)10LogM(1)(1/M1/W10Log10LogM 1a1aBest

N

1i

iNew

A

++−+

> ∑

=

/2)H(RW)10LogM(1MW10Log10LogM 1a1aBest

N

1i

iNew

A

+−−+

< ∑

=




Time

AbsoluteThreshold

PCPICH

Ec/No

ReportingRange

CPICH 2

CPICH 1

Event

1b

Figure 24

Event 1b: A PCPICH Leaves the Reporting Range

SC2804/S6/v1.1





5.5.4 Event 1c: A Non-Active PCPICH Becomes Better than an Active One

This report will be triggered when a PCPICH that is not in the active set is better thanthe worst PCPICH in the active set when the active set is full. This simply replacesone (the worst) PCPICH for another. A hysteresis known as the ‘replacementwindow’ is applied to this, meaning the new cell has to be better than the old by thishysteresis value.

5.5.5 Other Events

Event 1d: Change of Best CellThis report will be triggered when any PCPICH in the reporting range becomesbetter than the current serving cell plus a hysteresis value.

Event 1e A PCPICH becomes better than the absolute threshold plus a hysteresis value.

Event 1f A PCPICH becomes worse than the absolute threshold minus a hysteresis value.





Time

AbsoluteThreshold

CPICH 2

CPICH 3

CPICH 1

Hysteresis

ReplacementWindow

Hysteresis

Hysteresis

Event1e

Event1c

Event1d

Event1f

PCPICHEc/No

CPICH 4

Figure 25

Events 1c, 1d, 1e and 1f

SC2804/S6/v1.1





5.6 Optimizing Soft Handover Regions

There are a number of parameters that could be used to optimize the network inrespect of handovers, including:

• active set size

• reporting range

• absolute threshold

• hysteresis values

• time to trigger values

• offset values

5.6.1 Active Set Size

If the active set size is made too large it will result in unnecessary radio links beingestablished, resulting in more signalling being required to add more cells to theactive set. The soft handover margin will be eroded, resulting in increased UE andNode B transmit powers. The net result will be a reduction in downlink and uplinkcapacity. However, the impact of choosing too large a value can be controlled by the

other soft handover parameters controlling the addition of cells to the active set.

If the active set is made too small, frequent signalling and delayed handover willresult, degrading performance in both uplink and downlink directions. Consequently,that the UE and Node B power levels will need to increase giving more interferenceand reduced capacity.





Active set size

Preventsnecessary

SHO

Increased

TX power

DegradedBLER UL/DL

Increasedcall drop

rate

SHOparametersset correctly

?

Unnecessaryradio links

Increased

SHO

IncreasedTX power

ReducedUL/DL capacity

Too small Too large

Littleimpact

Yes No

Figure 26

Active Set Size

SC2804/S6/v1.1





5.6.2 The Reporting Range

The measurements on a cell must be within the reporting range before Event 1atakes place and the cell is subsequently added to the active set. The size of thereporting range is made relative to the serving cell and is influenced by a largenumber of parameters. A hysteresis value H1a along with a reporting range constantR1a define the Addition Window. Neighbour cells that enter the addition window willbe added to the active set.

The reporting range is also used to identify cells that should be removed from theactive set (Event 1b). A different hysteresis value, H1b, and the reporting rangeconstant R1b may be used to define a Drop Window. This would be used to removecells from the active set.

If a non-active PCPICH enters the reporting range when the active set is full and isfound to be better than an active one, Event 1c may take place. Here, the non-activePCPICH replaces the worst PCPICH. A hysteresis value can be applied, known asthe replacement window.





Time

AbsoluteThreshold

ReplacementWindow

PCPICHEc/No

DropWindow

AdditionWindow

Event1a

Figure 27

Reporting Range

SC2804/S6/v1.1





5.6.3 Addition Window

The addition window determines which cells are to be included in the active setrelative to the serving cell. If this window is too large the SHO area will be too large,resulting in too many soft handovers. The soft handover gain will diminish, resultingin increased transmit power levels, which will reduce uplink and downlink capacity.Downlink capacity may also suffer as a result of the availability of spreading codes.Exceeding the 30–40% SHO probability will also demand greater backhaul capacity.

Making the SHO area too small may result in frequent active set updates, which willplace a burden on signalling mechanisms. Fewer cells in SHO will reduce the SHOgain therefore demanding greater transmit power, which will reduce uplink anddownlink capacity.





Addition window

SHO regiontoo small

Frequentactive setupdates

Signallingburden

SHO regiontoo large

Increaseddemand onbackhaul

Too small Too large

Few cellsin SHO

Increasedtransmitpower

ReducedUL/DL

capacity

Too manySHO

Increasedtransmitpower

Reduced

UL/DLcapacity

Reduced DLcapacity

DiminishedSHO gain

Figure 28

Addition Window

SC2804/S6/v1.1





5.6.4 Drop Window

The drop window is set relative to the addition window, but the hysteresis values areset to make the drop window larger by a couple of dBs. If the window is set too largethen the wrong cells will be in the active set, making the soft handover regions larger.The net result of this will be increased transmit powers and reduced uplink anddownlink capacity.

If the drop window is too small there will be fewer cells in the active set, which willreduce the SHO gain and result in higher transmit powers and a reduction in uplinkand downlink capacity. Depending upon the terminal speed there may also befrequent handovers taking place, which will put a burden on signalling mechanisms.





Fewer cellsin active set

Reduced

SHO gain

Increasedtransmit power

Reduced UL/DLcapacity

Wrong cellsin active set

Large SHO

regions



Drop WindowToo small Too large

Figure 29

Drop Window

SC2804/S6/v1.1





5.6.5 Replacement Window

The replacement window is used when a new PCPICH is found and the active set isfull. The weakest PCPICH is replaced by the new PCPICH if it exceeds thereplacement window. If the replacement window is too large there will be fewer replacements, making the active set less optimal. This will result in increasedtransmit powers and ultimately a reduction in downlink and uplink capacity.

If the window is too small there will be excessive replacements and a ping-pongeffect will occur.





Excessivereplacements

Ping-pong

effects

Excessivesignallingburden

Fewer replacements

Non-

optimal



Replacement windowToo small Too large

Figure 30

Replacement Window

SC2804/S6/v1.1





5.6.6 Absolute Threshold

The absolute threshold guarantees the minimum quality for a cell. It is used withreporting Events 1e and 1f.

For Event 1e the trigger is defined by:

MNew ≤ T1e – H1e/2 for path loss

or

MNew ≥ T1e + H1e/2 for all other measurements

Where:

MNew is the measurement result of a cell that becomes better than anabsolute threshold

T1e is the absolute threshold

H1e is the hysteresis value

For Event 1f the trigger is defined by:

MNew ≥ T1f + H1f /2 for pathloss

or

MNew ≤ T1f – H1f /2 for all other measurements

Where:

MNew is the measurement result of the cell that becomes worse than theabsolute threshold

T1f is the absolute thresholdH1f is the hysteresis value





Time

AbsoluteThreshold

CPICH 3

CPICH 2

CPICH 1

Event

1e

Event

1f

PCPICH

Ec/No

Figure 31

Absolute Threshold

SC2804/S6/v1.1





5.6.7 Hysteresis Values

Hysteresis values can be used to limit the number of event-triggered reports. Thevalue of the hysteresis is given to the UE in the Measurement Control message.

In the case of Event 1d, where there is a change of best cell, this will not be reporteduntil the difference is equal to the hysteresis value. Choosing a large value for thehysteresis will make the change less likely. A smaller value would make the changemore likely, possibly resulting in a ping-pong effect.

The hysteresis values ranging from 0 to 7.5 dB in steps of 0.5 dB.

5.6.8 Time-to-Trigger

To minimize the number of signalling messages being transmitted by the UE; a time-to-trigger parameter can be given to the UE in the Measurement Control message.

The effect of this trigger is to ensure that a report is only triggered if the measuredresults are consistent and stable.

The time-to-trigger values are integer values of 0, 10, 20, 40, 60, 80, 100, 120, 160,

200, 240, 320, 640, 1280, 2560, and 5000 milliseconds.





Time

PCPICHEc/No

ReportingRange

Event 1aTime-to-Trigger

Figure 33

Time-to-Trigger

SC2804/S6/v1.1


Time

Event1d

PCPICHEc/No

Hysteresis

Figure 32

Hysteresis Values




5.6.9 Offset Values

For each cell in the monitored set an offset can be applied by the UE. This offsetvalue may be positive or negative and is given to the mobile for each cell concernedin the Measurement Control message.

By applying a positive offset the UE will send a measurement report as if thePCPICH Ec/No value was x dB greater than it actually is. This will allow thehandover algorithm to include the cell in the active set earlier than without an offset.

Applying a negative offset will cause the UE to subtract x dB before checking to seeif the cell is in the reported range. This will make the cell less attractive for handover purposes.

5.6.10 Forbidding a PCPICH to Affect the Reporting Range

The reporting range is defined as a function of all of the PCPICHs in the active set. If the weighting parameter, W, is set to zero, the reporting range is defined relative tothe best PCPICH. If there was a PCPICH in an area which was intermittently strong,i.e. became the best PCPICH intermittently, then the reporting range would becomeunstable. To prevent this occurring it is possible to bar the offending PCPICH from

the evaluation of the reporting range.





Time

PCPICHEc/No

ReportingRangeCPICH 1

CPICH 2

CPICH 3

Figure 35

Forbidding a PCPICH

SC2804/S6/v1.1


Time

Event1a

PCPICHEc/No

ReportingRangeCPICH 1

CPICH 2

CPICH 3

Event1b

Offset

Figure 34

Offset Values




5.7 Inter-Cell Synchronization

A Node B can support one or more cells. When a Node B creates three cells it isimportant that the radio frames transmitted over the air interface are not aligned. If they are, the synchronization channels in each cell will be aligned and causeexcessive noise. Instead, an offset known as T_cell is applied to each cell to stagger the transmission of the radio frames.

The parameter T_cell has a resolution of 256 chips with values between 0 and 9.

T_cell is applied to the Node B Frame Number (BFN) in the Node B to calculate thecell System Frame Number (SFN) as follows:

SFN = BFN adjusted with T_cell





Cell 1

Cell 2Cell 3

SCH

T_cell

T_cell

Cell 1

Cell 2

Cell 3

10 ms radio frame

SFN = BFN adjusted with T_cell

where T_cell = 0 9 x 256 chips

Figure 36

Inter-Cell Synchronization

SC2804/S6/v1.1





A hard handover involves changing frequency. It is a break-before-make process as

in GSM. Hard handovers require the UE to change from one UMTS frequency toanother, e.g. when moving between hierarchical cells. This is also known as an inter-frequency handover. Hard handovers are also performed when moving from a UMTSfrequency to a GSM frequency. This is known as an inter-system handover.

Hard handovers in UMTS are Mobile Evaluated Hand Overs (MEHO). The UE isrequired to perform measurements on neighbours and report measurement results.The reporting process may be periodic or based upon triggers. However, becausethe neighbours are on different frequencies and the UE is using CDMA on the air interface, it is necessary to use compressed mode.

6.1 Compressed Mode Measurements

The UE is able to perform a single measurement type in one transmission gappattern sequence. A transmission gap pattern sequence can consist of alternatingtransmission gap patterns 1 and 2. Each of these patterns in turn consists of one or two transmission gaps. The following parameters apply:

• Transmission Gap Starting Slot Number (TGSN)

• Transmission Gap Start Distance (TGD)• Transmission Gap Length 1 (TGL1)

• Transmission Gap Length 2 (TGL2

• Transmission Gap Pattern Length (TGPL)

• Transmission Gap Pattern Repetition Count (TGPRC)

The TGSN is the slot number of the start of the first transmission gap in the firsttransmission gap pattern. The time to the start of the second transmission gap isgiven by the TGD. The duration of each transmission gap is given by TGL1 and

TGL2. Note that it is possible that only one transmission gap will be requested. Thelength of the pattern containing the two transmission gaps is defined by TGPL1. Notethat a second pattern length, TGLP2, may also be used. The sequence defined byTGPL1 and TGPL2 continues for a total number of frames defined by TGPRC.

6 HARD HANDOVERS







6.1.1 FDD Inter-Frequency Measurements

In order for the UE to perform measurements on other FDD carriers the RNC mustprovide a transmission gap pattern sequence using the parameters TGL1, TGL2,TGD and Max TGPL.








6.1.2 GSM Carrier RSSI

For a UE to be able to perform RSSI measurements on a GSM carrier the RNC mustprovide a transmission gap pattern sequence using the parameters TGL1, TGL2,and TGD. See Figure 39a.

The Transmission Gap Length will dictate how many RSSI samples can be taken ona GSM carrier. To meet the measurement accuracy requirements stated in 3GPP TS45.0081 the measurement time should allow the UE to take three RSSI samples per GSM carrier in the monitored set. This will require a TGL value no smaller than fiveslots. See Figure 39b.

Figure 39c shows the combinations of TGL1, TGL2 and TGD, which will be used if the UE is also required to perform BSIC verification.

1 3GPP TS 45.008 Radio Subsystem Link Control.




TGL1 (Slots) TGL2 (Slots) TGD (Slots)

3 - undefined

4 - undefined

5 - undefined

7 - undefined

10 - undefined

14 - undefined

3 3 15…269

4 4 15…269

5 5 15…269

7 7 15…26910 10 15…269

14 14 15…269

Figure 39a

RSSI Measurements

SC2804/S6/v1.1


TGL1 (Slots) TGL2 (Slots) TGD (Slots)

5 - undefined

7 - undefined

10 - undefined

14 - undefined

5 5 15…269

7 7 15…269

10 10 15…269

14 14 15…269

Figure 39c

BSIC Verification

TGL Number of GSM carrier RSSI samples in each gap.

3 1

4 2

5 3

7 6

10 10

14 15

Figure 39b

RSSI Samples Per Carrier




6.2 Neighbour Cells for Hard Handover

In the CELL_DCH and CELL_FACH states the UE can monitor up to 32 inter-frequency cells, including FDD cells using up to two FDD carriers and, depending onUE capability, 32 GSM cells using up to 32 carriers.

In the CELL_DCH state, when the compressed mode of operation is supported, theUE continuously measures on identified inter-frequency cells and searches for newcells as indicated in the Measurement Control message.

The neighbour lists are broadcast as part of system information, but a UE inconnected mode can receive details using dedicated signalling via the DCCH.


• Global RNC Id (MCC + MNC and RNC Id)

• Cell Identifier (CI)

• Location Area Code (LAC)

• Routing Area Code (RAC)

• UARFCN

• Scrambling code for PCPICH


• Cell Global Identity (CGI)

• BCCH frequency

• Base Station Identity Code (BSIC)





In CELL_DCH and CELL_FACH states the UE can monitor up to:



32 inter-frequency cells, including FDD cells using up to two FDD carriers

depending on UE capability, 32 GSM cells using up to 32 carriers

Global RNC Id (MCC + MNC and RNC Id)

Cell Identifier (CI)

Location Area Code (LAC)

Routing Area Code (RAC)

UARFCN

scrambling code for PCPICH

Cell Glocal Identity (CGI)

BCCH frequency

Base Station Identity Code (BSIC)

Figure 40

Neighbour Cells for Hard Handover

SC2804/S6/v1.1





6.3 Hard Handover Parameters and Triggers

6.3.1 Inter-Frequency Handovers

Quality estimates are performed on interfrequency measurements according to thefollowing:

Where:

Qcarrier j is the estimated quality of the active set on frequency j.

Mcarrier j is the estimated quality of the active set on frequency j.

Mi j is a measurement result of cell i in the active set on frequency j.

N A j is the number of cells in the active set on frequency j.

MBest j is the measurement result of the strongest cell in the active set onfrequency j.

W j is a parameter sent from UTRAN to UE and used for frequency j.

H is the hysteresis parameter.

The measurement control message notifies the UE about which events to use totrigger a measurement report. The triggers are:

• Event 2a – change of best frequency

• Event 2b – estimated quality of a currently-used frequency is below a threshold,and the estimated quality of a non-used frequency is above a threshold

• Event 2c – the estimated quality of a non-used frequency is above a threshold

• Event 2d – the estimated quality of the currently-used frequency is below athreshold

• Event 2e – the estimated quality of a non-used frequency is below a threshold

• Event 2f – the estimated quality of the currently-used frequency is above athreshold


H)10LogMW(1M10LogW10LogMQ jBest j

N

1i

ji j jcarrier jcarrier

j A

−−+

== ∑

=




Event 2a – change of best frequency

Event 2b – estimated quality of a currently-used frequency is below a threshold,and the estimated quality of a non-used frequency is above a threshold

Event 2c – the estimated quality of a non-used frequency is above a threshold

Event 2d – the estimated quality of the currently-used frequency is below a

threshold

Event 2e – the estimated quality of a non-used frequency is below a threshold

Event 2f – the estimated quality of the currently-used frequency is above athreshold

Figure 41

Inter-Frequency Events

SC2804/S6/v1.1





Event 2a

If the quality estimate of a non-used frequency exceeds the quality estimate for thecurrently-used cell and Event 2a has been requested by the RNC, a report will betriggered. A hysteresis value and time-to-trigger will also apply to this event.

Event 2bThe UE will be given two threshold values in the Measurement Control message.There is a threshold for the used frequency and a threshold for the non-usedfrequency. If the estimated quality on a used frequency falls below its definedthreshold and the estimated quality for a non-used frequency exceeds its definedthreshold a report is triggered.

Event 2cIf ordered by the RNC, the UE will send a report if the estimated quality on a non-used frequency exceeds a predefined threshold. A hysteresis value and time-to-trigger also apply to this event.

Event 2dWhen ordered by the RNC, the UE will report when the estimated quality on acurrently-used frequency falls below a predefined threshold. A hysteresis and time-to-trigger applies to this event.

Event 2eThe UE will send a report to the RNC when the estimated quality on a non-usedfrequency is below a threshold after applying a hysteresis and time-to-trigger value

Event 2f The UE sends a report to the RNC when the estimated quality of a currently-usedfrequency is above a threshold taking into account hysteresis and time-to-trigger values.





Time

EstimatedQuality

2f

ThresholdCell 1

ThresholdCell 2

Cell 1

2c 2b + 2d

2a

2e

Cell 2

Figure 42

Inter-Frequency Triggers

SC2804/S6/v1.1





6.3.2 Inter-System Handovers

Where:

QUTRAN is the estimated quality of the active set on the currently-used UTRANfrequency.MUTRAN is the estimated quality of the active set on currently-used UTRANfrequency expressed in a unit other than that used on the current UTRAN.

Mi is a measurement result of cell i in the active set.N A is the number of cells in the active set.

MBest is the measurement result of the strongest cell in the active set.


The measurement control message notifies the UE about which events to use totrigger a measurement report. The triggers are:

• Event 3a – the estimated quality of the currently-used UTRAN frequency is

below a certain threshold and the estimated quality of the other system isabove a certain threshold

• Event 3b – the estimated quality of the other system is below a certainthreshold

• Event 3c – the estimated quality of the other system is above a certainthreshold

• Event 3d – change of best cell in other system


Best

N

1iiUTRANUTRAN W)10LogM(1MW10Log10LogMQ

A

−+

== ∑

=




Event 3a – the estimated quality of the currently-used UTRAN frequency is belowa certain threshold and the estimated quality of the other system is above a certainthreshold

Event 3b – the estimated quality of the other system is below a certain threshold

Event 3c – the estimated quality of the other system is above a certain threshold

Event 3d – change of best cell in other system

Figure 43

Inter-System Events

SC2804/S6/v1.1





Event 3a

If the estimated quality on the currently used frequency falls below the ThresholdOwn System and the estimated quality of the other system is above the ThresholdOther System a report will be triggered. Hysteresis and time-to-trigger values areapplied to both sets of comparisons.

Event 3bIf the estimated quality on the other system falls below the Threshold Other Systema report will be triggered. Hysteresis and time-to-trigger values are also applied.

Event 3cWhen the estimated quality on the other system exceeds the Threshold Other System a report will be triggered. Hysteresis and time-to-trigger values are applied.

Event 3dIf the quality estimates for a cell in the other system exceed the quality estimate for the best cell in the other system a report will be triggered. Hysteresis and time-to-trigger values are also applied.





3c 3b

Time

EstimatedQuality Own Cell

3a

ThresholdOwn

System

Threshold

Other System

Other Cell

Figure 44

Inter-System Triggers

SC2804/S6/v1.1









SECTION 7

UMTS FEATURES AND TECHNIQUES

i









1 Current Optional Features 7.11.1 Introduction 7.11.2 Site Selection Diversity Transmit (SSDT) 7.11.3 Transmit Diversity Options 7.31.4 Open Loop Mode 7.31.5 Closed Loop Mode 7.7

1.6 Impact of Transmit Diversity 7.71.7 Multi-User Detection (MUD) 7.91.8 Advanced Antennas 7.11

2 Future Optimal Features 7.172.1 Introduction 7.172.2 High Speed Downlink Packet Access (HSDPA) 7.172.3 New Transport and Physical Channels 7.172.4 Channel Adaptation 7.172.5 Implementation of the HS-DSCH 7.192.6 HARQ and Adaptive Channels 7.21

2.7 Adaptive Modulation 7.212.8 Multiple Input Multiple Output (MIMO) Antennas 7.232.9 Interworking with Wireless LANs (WLAN) 7.25

CONTENTS

iii











1.1 Introduction

A number of features may be employed in UMTS to optimize the network. All of thefeatures are mandatory for UEs but are optional on the network side. These optionalfeatures include:

• Site Selection Diversity Transmit (SSDT)

• transmit diversity

• Multi-User Detection (MUD)

• smart antennas

1.2 Site Selection Diversity Transmit (SSDT)

Site Selection Diversity Transmit (SSDT) is a technique that can be applied when aUE is engaged in a soft handover to reduce downlink interference.

The principle of SSDT is that the UE will dynamically indicate the best cell from itscurrent active set based on the downlink reception level measurement of the CPICH.Each cell in the active set is given a temporary ID within the set; the UE indicates theprimary cell using part of the Feedback Information (FBI) field in the uplink DPCCH.The nominated primary cell then transmits both DPDCH and DPCCH. All other cellsin the active set are selected as non-primary and only transmit DPCCH.

The UE continues to monitor the DPCCH for all cells in the active set. If a non-primary cell is judged to be better than the nominated primary, the UE indicates thischange to the RNC. The new primary will then begin to transmit DPDCH and the oldprimary will continue with the transmission of DPCCH only.

The lack of downlink DPDCHs will reduce the soft handover gain experienced by theUE. There will be no loss of gain in the uplink direction because all Node Bs engaged

in soft handover will be listening to both DPDCH and DPCCH from the UE.

It has been suggested that there may be a loss of capacity as a result of SSDT. It hasbeen shown through simulations that although the capacity gain is large for high-bit-rate services, it is small for low-bit-rate services such as speech. In order to improvethe performance of SSDT, the use of enhanced SSDT has been discussed in RANWG1.

1 CURRENT OPTIONAL FEATURES





SRNC

UE nominatesNode B

as Primary

Iub Iub

IubDPCCH only

DPCCH only

DPDCH/DPCCH

1

2

3

1

Figure 1

SSDT

SC2804/S7/v1.1





1.3 Transmit Diversity Options

There is no requirement for a 3G handset to support receive diversity, but it ispossible to improve downlink performance using transmit diversity techniques.

There are two ways of implementing transmit diversity: closed loop, where the UEreports the performance back to the Node B; and open loop, where no reporting isnecessary.

1.4 Open Loop Mode

In open loop mode there are two types of transmit diversity available: Space TimeTransmit Diversity (STTD) and Time Switched Transmit Diversity (TSTD).

1.4.1 Space Time Transmit Diversity (STTD)

In STTD two signals are transmitted from two transmission antennas simultaneously,as illustrated in Figure 2. The two signals are received by the UE on the samepropagation paths but with uncorrelated fading characteristics. This provides spacediversity.

Time diversity is achieved by passing the data through an orthogonal block-encodingprocess prior to transmission. Due to the orthogonality of the block-encoding schemeover a sequence of 4 bits (2-Phase Quadrature Phase Shift Keying (2-QPSK))symbols in the downlink) it is possible for the UE to separate the two signalcomponents from the separate antennas (so long as the radio path remains timeinvariant over an interval corresponding to 2-QPSK symbols) and perform optimumcombining.

In addition to data and control signals, pilot signals can also be transmitted usingSTTD. For a detailed description of STTD encoding refer to 3GPP TS 25.211.





Antenna1

Antenna2

Channel bits

STTD orthogonallyencoded bits over

2 QPSK symbols

Antenna1

Antenna2

b3b2b1b0

b3b2b1b0

–b 1b0b3 –b 2

Figure 2

Transmit Diversity

SC2804/S7/v1.1





1.4.2 Time Switched Transmit Diversity (TSTD)

TSTD can be applied to the SCH. With reference to Figure 3, in even-number slotsboth the Primary SCH (P-SCH) and the Secondary SCH (S-SCH) are transmitted onantenna one, and in odd-number slots both P-SCH and S-SCH are transmitted onantenna two.





Antenna

1

Antenna

2

Slot 0 Slot 1 Slot 2 Slot 14

Slot 0 Slot 1 Slot 2 Slot 14

S-SCH

P-SCH

S-SCH

S-SCH

P-SCH

P-SCH

Gp

Gs

G – Gain Control

S-SCH

P-SCH

TX off

TX off

TX off

TX off

TX off

TX off

TX off

TX off

S-SCH

P-SCH

Figure 3

TSTD

SC2804/S7/v1.1





1.5 Closed Loop Mode

Closed loop mode is only applicable for use in dedicated channels. In closed loopmode the UE monitors the CPICH transmissions from the serving cell to assess theperformance of transmit diversity. It then uses defined algorithms to calculate theoptimal settings required by the Node B for best link performance. The requirementsare conveyed to the Node B in the D field of the FBI bits within the DPCCH. Thereare two closed loop modes: closed loop mode 1 and closed loop mode 2.

With reference to Figure 4, the DPCH for transmission is applied to both transmitantenna branches and weighted with antenna-specific weighting factors W1 and W2.

In closed loop mode 1 the phase of one antenna is adjusted relative to the other withan accuracy of 1 bit per timeslot. In closed loop mode 2 both the relative phase andamplitude are adjusted with an accuracy of 1 bit.

1.6 Impact of Transmit Diversity

The additional multipaths created by transmit diversity may result in the loss of downlink orthogonality for the spreading codes. This would degrade the downlinkperformance, in particular for terminals moving at speed through a macro cellular

environment. Simulation results have shown that the greatest benefit is achievedwhen transmit diversity is introduced into cells that have little multipath, such asmicro cells.

Both open loop and closed loop transmit diversity offer the benefit of a reducedEb/No requirement. Simulations of a 12.2 kbit/s service with a BLER target of 1%have shown a 0.5 to 3 dB reduction in the Eb/No requirement for open loop mode.But it depends on terminal speed and environment. This can be improved by afurther 0.5 dB using closed loop mode 1.

These improvements impact upon downlink system capacity and the downlink

coverage area. Improving the downlink coverage area is particularly important inmicro cells, where the Node B transmit power is relatively small. Simulations havealso shown that capacity gains of up to 70% may be achieved in a micro cell usingclosed loop mode 1.





CPICH

DPCH

Weight(W2)

Weight(W1)

Antenna 1

Antenna 2

CPICH

DPCCH

Weight Generation

FBI D Field Extraction

Figure 4

Closed Loop Mode

SC2804/S7/v1.1





1.7 Multi-User Detection (MUD)

Multi-User Detection (MUD) is a form of noise suppression process. In active noisesuppression systems, the ambient noise is sampled, inverted and transmitted backinto the environment in real time to cancel the noise.

In a CDMA system, the noise on a wanted channel is largely the interference of other users in the cell. Although noise-like in nature, it is determinable as they are alsobeing decoded at the base station for the other users in their channels.

A conceptual diagram of a base station with MUD is shown in Figure 5.

The outputs of rake receivers 2 and 3 are weighted and fed back to receiver 1.





etc.

User 1

User 2

User 3

RakeReceiver

WeightedCorrectionRF down

conversionand A to Dconversion

RakeReceiver

RakeReceiver

Figure 5

Conceptual MUD Receiver

SC2804/S7/v1.1





1.8 Advanced Antennas

Traditional cellular antennas have fixed radiation and reception patterns and the gainof the antenna is proportional to its size. Research into advanced antennas has beendriven by the need for range extension and controlling interference. Controllinginterference will bring about capacity gains and together with range extensionimprovements in coverage will be possible. The technology behind advancedantennas includes switched beam arrays and adaptive (smart) antennas.

1.8.1 Switched Beam Arrays

A switched beam array comprises a phased array antenna and some logic to switcha radio connection from one beam to the next. A phased array antenna comprises anumber of fixed antenna elements into which power is delivered with different relativephases. The signals are summed coherently in a specific Direction of Arrival (DoA).

In the uniform phased array antenna the phase shift relative to one antenna elementincreases linearly from element to element. The phase shift is a function of theelement spacing, d, DoA, θ, and wavelength, λ . The simplest switched beam arrayuses the Butler Matrix to define the phase shifts associated with each beam.

A user would be switched from beam to beam very much like moving from cell to cellusing conventional antennas. Allocated to each beam there will be a secondaryCPICH to serve as a phase reference and to be used for measurement purposes.Based upon the mobile’s reported measurements the RNC can switch the mobilefrom beam to beam by performing handovers. In addition, secondary cell scramblingcodes can be allocated to the individual beams, allowing the reuse of spreadingcodes.






7.13 © Wray Castle Limited

Beam1 2 3 4

1 0º –135º –270º –405º

2 0º –45º –90º –135º3 0º 45º 90º 135º

4 0º 135º 270º 405º

Antenna Elements

Figure 7

Butler Matrix

SC2804/S7/v1.1





–60 –40 –20 0 20 40 60

Conventionalcell

degrees

Figure 8

Beam Pattern

SC2804/S7/v1.1





1.8.2 Adaptive (Smart) Antennas

Adaptive (or smart) antennas use the same phased array as switched beamantennas but require more sophisticated logic. This logic will provide Spatial FilteringInterference Rejection (SFIR) and Spatial Division Multiple Access (SDMA).

SFIR is a process where the array of elements provides a composite coveragepattern with a null in the direction of an interferer and dynamically steering the null asthe interference moves.

SDMA includes the SFIR technique but is also able to steer the lobe serving amobile and dynamically adjust the power in that lobe. Using multiple lobes will allowa number of mobiles to be served in the same area.

Smart antennas increase the complexity of a system and may not support transmitdiversity. There will also be a major impact on the radio resource managementfunction in the RNC. Consequently, smart antennas may not find an application in3G.

1.8.3 Impact of Advanced Antennas

Simulation results comparing a four-beam array with polarization diversity withconventional two antenna diversity are very favourable. In the uplink direction thereduction in required Eb/No ranged from 5–6.5 dB depending upon terminal speedand operating environment.

In the downlink direction, comparison of a four-beam array with a single transmitantenna in a macro cell gave a 4.5 dB reduction in the required Eb/No value.





a)

Azimuth and radiated powerof beam(s) may be

dynamically adjusted toaccount for traffic distribution

and interference sources

Interference

B

b)

B

Interference

Figure 9

Adaptive (Smart) Antennas

SC2804/S7/v1.1





2.1 Introduction

UMTS is evolving, and subsequent releases of the 3GPP specification, namelyReleases 4, 5 and 6, will bring new features which may be used to optimize networkperformance. Some of these are:

• High Speed Downlink Packet Access (HSDPA)

• Hybrid ARQ (HARQ) and adaptive channels

• Multiple Input Multiple Output (MIMO) antennas

• interworking with Wireless LANs (WLAN)

2.2 High Speed Downlink Packet Access (HSDPA)

A number of advanced techniques have been put forward to achieve high bit rateson the air interface under the general heading of HSDPA. Downlink data rates of 8–10 Mbit/s are possible, increasing to 20 Mbit/s in the future. Details of thesetechniques are included in Release 5 and Release 6 of the UMTS specifications.

2.3 New Transport and Physical Channels

New transport channels have been defined: the High Speed Downlink SharedChannel (HS-DSCH) and the High Speed Shared Control Channel (HS-SCCH).

2.4 Channel Adaptation

Adaptive Modulation and Coding (AMC) is a mechanism whereby the modulationschemes of four-state Quadrature Phase Shift Keying (QPSK) and 16-stateQuadrature Amplitude Modulation (16QAM) can be chosen dynamically according to

the suitability of the radio environment. This also includes adjusting the ForwardError Correction (FEC) and puncturing rates. Once the FEC checks have beenmade, Hybrid Automatic Repeat Request (ARQ) provides soft combining of allretransmissions.

Release 6 will include Multiple Input Multiple Output (MIMO), whereby severaltransmit antennas can be employed at the base station along with several antennasat the receiver. This allows downlink bit rates to achieve 20 Mbit/s.

2 FUTURE OPTIMAL FEATURES





High Speed Downlink Packet Access (HSDPA)

HS-DSCH DL Transport

HS-SCCH DL Physical

HS-PDSCH DL Physical

HS-DPCCH UL Physical

8–10 Mbit/s

20 Mbit/sQPSK

16QAM

HARQ

MIMO

Figure 10

HSDPA

SC2804/S7/v1.1





2.5 Implementation of the HS-DSCH

The HS-DSCH is implemented at the MAC layer. Under normal circumstancesdedicated and shared transport channels would be implemented in MAC-d andMAC-c/sh located in the RNC. However, because of the need for high-speed datatransfer with error correction, a new MAC entity, MAC-hs, will be implemented in theNode B. This will eliminate the need for retransmission of erroneous data over theIub interface, reducing delays.

MAC-hs is responsible for handling the data transmitted on the HS-DSCH. Additionally, it is responsible for the management of the physical resources allocatedto HSDPA. MAC-hs is composed of four different functional entities:

• flow control

• scheduling/priority handling

• HARQ

• Transport Format Resource Combination (TFRC) selection

Flow control is used to reduce discarded and retransmitted data as a result of theHS-DSCH congestion.

The scheduling/priority handling manages the HS-DSCH resources between HARQentities and data flows, according to their priority.

The HARQ handles the control of errors whilst TFRC is responsible for the selectionof the appropriate transport format for the data to be transmitted.





Node B Iub RNC

MAC-hs

MAC-c/sh

Scheduling/priority handling

HARQ

TFRC selection

HS-DSCH

DTCH

MAC-d

DTCH

Figure 11

Implementation of the HS-DSCH

SC2804/S7/v1.1





2.6 HARQ and Adaptive Channels

In conventional ARQ schemes, frame errors at the receiving end are detected using aCyclic Redundancy Checksum (CRC). If a frame is received in error a NegativeAcknowledgemet (NACK) is returned prompting a retransmission of the erroneousframe. Otherwise, an Acknowlegement (ACK) positively acknowledges the receivedframe.

User data and the CRC bits may be additionally encoded by an error correcting code,which increases the probability of successful transmission. Such schemes are knownas Hybrid ARQ (HARQ) schemes.

A measure of ARQ protocol efficiency is throughput, defined as the average numberof user bits accepted at the receiving end in a given time. The more redundant bitstransmitted, the lower the efficiency.

In mobile environments the Incremental Redundancy (IR) HARQ scheme exhibitshigher throughput efficiency by adapting the error correcting code redundancy todifferent channel conditions. A block of user data is sent with a CRC and parity bits. Ifthe CRC checksum fails in the receiver a NACK is returned to the transmitter, whichtransmits additional parity bits only. These bits are combined with the first in a secondattempt to correct the error. If the CRC checksum fails again, additional parity bits are

transmitted until the receiver can decode the information successfully. By onlyretransmitting parity bits the throughput will be greatly improved.

2.7 Adaptive Modulation

High-order modulation schemes such as 16QAM provide high spectral efficiency interms of bit/s/Hz compared to QPSK, and can provide much higher peak data rates.However, the air interface is a hostile environment and 16QAM is not as robust asQPSK. One way in which 16QAM could be used advantageously would be byallocating proportionally more power to that channel.





Transmitter

CRC fails

CRC fails

CRC OK

Receiver

Data and parity and CRC

ACK

Data and parity and CRC

NACK

Parity

NACK

Parity

ACK

Figure 12

Hybrid ARQ

SC2804/S7/v1.1





2.8 Multiple Input Multiple Output (MIMO) Antennas

MIMO is a technique for increasing data rates over a radio interface. In 1998, BellLabs demonstrated a new technique to greatly increase the capacity of a radio linkwhich has become known as Multiple Input Multiple Output (MIMO). Researchers atBell Labs used temporal and multi-antenna spatial diversity techniques in anarchitecture called BLAST (Bell labs LAyered Space-Time).

In conventional wireless systems, multipath propagation is problematic becausecomponents arrive at the receiver at slightly different times, giving rise to fast fadingand time dispersion. However, MIMO techniques exploit multipath to enhanceperformance by treating the multiple components as separate parallel subchannels.

This is achieved by splitting a single user’s data stream into multiple substreams andusing an array of antennas to simultaneously transmit the parallel substreams. Sincethe user’s data is being sent in parallel via multiple antennas, the effective bit rate isincreased roughly in proportion to the number of antennas.

At the receiver, an array of antennas is used to pick up the multiple substreams andtheir multipath components. Each antenna ‘sees’ all of the transmitted substreamssuperimposed. However, if there is sufficient multipath scattering, the multiplesubstreams are all scattered slightly differently, since they originate from different

transmit antennas that are located at different points in space. Using sophisticatedsignal processing, these slight differences in scattering allow the substreams to beidentified and recovered.

The signal processing algorithms used at the receiver are central to the technique. Atthe bank of receiving antennas, high-speed signal processors look at the signalsfrom all the receiver antennas simultaneously, first extracting the strongestsubstream then proceeding with the remaining weaker signals.

It is anticipated that using MIMO within the UTRAN will allow transmitted bit rates tobe increased five fold. Potentially, a downlink bit rate of 20 Mbit/s could be achieved

if MIMO was used with HSDPA.





Transmitter Receiver

Figure 13

MIMO

SC2804/S7/v1.1





2.9 Interworking with Wireless LANs (WLAN)

Wireless LANs (WLAN) offer broadband-style access to the public Internet. MostWLANs conform to IEEE 802.11b or 802.11g standards. 802.11b has existed for anumber of years and there are numerous of ‘WiFi® hotspots’ in public places such asfast food restaurants, bars and shops. 802.11b offers Internet access at bit rates upto 11 Mbit/s. The more recent 802.11g standard is backwards-compatible with802.11b, but offers bit rates up to 54 Mbit/s. WLANs are commonly used oncorporate intranets to simplify network design and implementation.

The purpose of interworking UMTS with WLANs is to extend the UMTS services andfunctionality to the WLAN environment. The interworking system will provide bearer services for connecting a UMTS subscriber via a WLAN to access IP-based servicescompatible with those offered via the packet-switched domain.

The interworking specification (3GPP 23.234) defines two procedures for the UMTSsystem. The first is WLAN Access, Authentication and Authorization, which will allowaccess to the WLAN and the locally connected IP network. Authentication and Authorization is done by the UMTS system and access to the locally connectednetwork is known as WLAN Direct IP Access. Secondly, WLAN 3GPP IP Access willallow WLAN UEs to establish a connection to 3GPP IP-based services or theInternet via the UMTS network.

2.9.1 Interworking Network Elements

The WLAN UE is the user equipment equipped with a UMTS Integrated ServiceCard (UICC) card. The UE may be capable of WLAN access only, or both WLAN andUMTS operation.

The 3GPP AAA server deals with Authentication, Authorization and Accounting for individual WLAN UEs accessing the system. The 3GPP AAA Server will beimplemented in the 3GPP network as a proxy server.

The packet data gateway will allow services on the 3GPP packet-switched networkto be accessed by the WLAN UE.





WLAN UE

Internet/intranet

3GPP network

3GPP AAAServer

WLAN3GPP IPaccess

802.11

WLANPacketData

Gateway

3GPP PSservices

and Internetaccess

Figure 14

Interworking with Wireless LANs

SC2804/S7/v1.1









INTRODUCTION TO UMTS OPTIMIZATION

GLOSSARY OF TERMS

i

Glossary of Terms



SC2804/Glossary/v1.1II © Wray Castle Limited

Glossary of Terms



SC2804/Glossary/v1.1 G.1© Wray Castle Limited

Glossary of Terms

2G Second Generation

2QPSK 2-state Quadrature Phase Shift Keying3G Third Generation3GPP 3rd Generation Partnership Project4QPSK 4-state Quadrature Phase Shift Keying16QAM 16-state Quadrature Amplitude Modulation

AAA Authentication, Authorization and Accounting ACI Adjacent Channel Interference ACK Acknowledgement ACLR Adjacent Channel Leakage Ratio AI Acquisition Indicator AICH Acquisition Indicator Channel AM Acknowledged Mode AMC Adaptive Modulation and Coding AMR Adaptive Multi Rate AP-AICH Access Preamble Acquisition Indicator Channel ARQ Automatic Repeat Request AS Access Stratum

BCCH Broadcast Control ChannelBCH Broadcast Channel

BFN Node B Frame Number BLAST Bell labs LAyered Space-Time ArchitectureBLER Block Error RateBSIC Base Station Identity CodeBTS Base Transceiver Station

CAC Call Admission ControlCCCH Common Control ChannelCD/CA-ICH Collision Detection/Channel Assignment Indicator ChannelCDMA Code Division Multiple AccessCGI Cell Global Identity

CI Cell Identifier CPCH Common Packet ChannelCPICH Common Pilot ChannelCRC Cyclic Redundancy ChecksumCRNC Controlling Radio Network Controller CSICH CPICH Status Indicator ChannelCTCH Common Traffic ChannelCW Carrier Wave



SC2804/Glossary/v1.1G.2 © Wray Castle Limited

DCCH Dedicated Control Channel

DCH Dedicated ChannelDL DownlinkDoA Direction of ArrivalDPCCH Dedicated Physical Control ChannelDPCH Dedicated Physical ChannelDPDCH Dedicated Physical Data ChannelDRX Discontinuous ReceptionDSCH Downlink Shared ChannelDTCH Dedicated Traffic Channel

EFR Enhanced Full RateEIRP Effective Isotropic Radiated Power

FACH Forward Access ChannelFBI Feedback InformationFDD Frequency Division DuplexFEC Forward Error Correction

GPS Global Positioning SystemGSM Global System for Mobile Communications

HARQ Hybrid Automatic Repeat RequestHCS Hierarchical Cell StructureHSDPA High Speed Downlink Packet AccessHS-DPCCH High Speed Dedicated Physical Control ChannelHS-DSCH High Speed Downlink Shared ChannelHS-PDSCH High Speed Physical Downlink Shared ChannelHS-SCCH High Speed Shared Control Channel

IEEE Institute of Electrical and Electronics EngineersIM Interference MarginIP Internet Protocol

IPDL Idle Period DownlinkIS Interim StandardIR Incremental Redundancy

KPI Key Performance Indicator

LAC Location Area CodeLCS Location ServicesLLC Logical Link ControlLMU Location Management Unit

LNA Low Noise Amplifier LOP Line of Position

Glossary of Terms



SC2804/Glossary/v1.1 G.3© Wray Castle Limited

MAC Medium Access Control

MCL Minimum Coupling LossMEHO Mobile Evaluated Hand Over MHA Mast Head Amplifier MIMO Multiple Input Multiple OutputMSC Mobile-services Switching CentreMUD Multi-User Detection

NACK Negative AcknowledgementNAS Non-Access StratumNF Noise FigureNMC Network Management CentreNRT Non-Real Time

O&M Operations and MaintenanceOMC Operations and Maintenance CentreOTDOA Observed Time Difference of ArrivalOTSR Omni Transmit Sector ReceiveOVSF Orthogonal Variable Spreading Factor

PA Power Amplifier PCCH Paging Control Channel

PCCPCH Primary Common Control Physical ChannelPCF Position Calculation FunctionPCH Paging ChannelPCPCH Physical Common Packet ChannelPDA Personal Digital AssistantPDC Personal Digital Cellular PDF Probability Distribution FunctionPDSCH Physical Downlink Shared ChannelPI Paging Indicator PICH Paging Indicator ChannelPLMN Public Land Mobile Network

PO Power OffsetPRACH Physical Random Access ChannelP-SCH Primary Synchronization Channel

QoS Quality of Service

Glossary of Terms



SC2804/Glossary/v1.1G.4 © Wray Castle Limited

RAB Radio Access Bearer

RAC Routing Area CodeRACH Random Access ChannelRAT Radio Access TechnologyRL Radio LinkRLC Radio Link ControlRNC Radio Network Controller RNS Radio Network SubsystemRRC Radio Resource ControlRSCP Received Signal Code Power RSSI Received Signal Strength IndicationRT Real TimeRTD Real Time DifferenceRTT Round Trip Time

SCCPCH Secondary Common Control Physical ChannelSCH Synchronization ChannelSDMA Spatial Division Multiple AccessSFIR Spatial Filtering Interference RejectionSFN System Frame Number SHO Soft Handover SID Silence Description

SIR Signal to InterferenceSRNC Serving Radio Network Controller S-SCH Secondary Synchronization ChannelSSDT Site Selection Diversity TransmitSTTD Space Time Transmit Diversity

TDD Time Division DuplexTFCI Transport Format Combination Indicator TFRC Transport Format Resource CombinationTGD Transmission Gap start DistanceTGL Transmission Gap Length

TGPL Transmission Gap Pattern LengthTGPRC Transmission Gap Pattern Repetition CountTGSN Transmission Gap Starting Slot Number TMA Tower Mounted Amplifier TPC Transmit Power ControlTSTD Time Switched Transmit Diversity

Glossary of Terms



Introduction to UMTS Optimization

Documents