8/13/2019 Introduction to UMTS Optimization http://slidepdf.com/reader/full/introduction-to-umts-optimization 1/429 Introduction to UMTS Optimization Course 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
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
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
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.
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.
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.
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.
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.
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
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.
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.
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.
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.
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.
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.
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.
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.
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.
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
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.
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.
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
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
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).
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.
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
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.
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
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.
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%.
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.
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.
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.
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:
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.
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.
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.
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.
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).
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.
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.
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.
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.
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.
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.
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.
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
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.
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
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.
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.
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.
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.
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.
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.
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.
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.
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.
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.
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.
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.
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.
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.
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.
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.
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.
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.
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.
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.
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
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
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.
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.
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.
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.
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.
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.
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.
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.
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.
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.
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.
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).
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
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.
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.
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.
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
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.
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.
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
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.
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).
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).
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.
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.
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).
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.
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.
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.
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.
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.
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.
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.
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.
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.
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.
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.
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.
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.
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.
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.
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.
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?
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).
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.
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.
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.
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.
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
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.
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.
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.
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.
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:
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.
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.
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
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.
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.
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:
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.
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
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.
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
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.
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.
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.
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
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
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.
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
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:
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
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.
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.
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.
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.
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.
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.
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.
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
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:
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.
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.
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.
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.
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 PCPICH
For a GSM cell the following information will be required:
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
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.
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.
W is a parameter sent from the UTRAN to the UE.
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 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
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.
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
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.
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.
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.
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.
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.
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.
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.
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.
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.
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.
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
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.
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
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
MCL Minimum Coupling LossMEHO Mobile Evaluated Hand Over MHA Mast Head Amplifier MIMO Multiple Input Multiple OutputMSC Mobile-services Switching CentreMUD Multi-User Detection
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
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