Presentation-IIR04-Tutorial-A Systematic Approach for Wcdma Design

8/8/2019 Presentation-IIR04-Tutorial-A Systematic Approach for Wcdma Design

1/65


2/65

2

AsystematicapproachforUMTSRAND

imensioning

Dr.NicolasCotanis

September 20, 2004

Proprietary & Confidential

Topics to address

Introduction UMTS specific design inputs

Traffic layer

RAN pre-design Static network simulation (design)

Design optimization Network optimization


3/65

3


imensioning

Dr.NicolasCotanis

September 20, 2004


Introduction UMTS specific design inputsUMTS specific design inputsUMTS specific design inputs

Traffic layerTraffic layerTraffic layer

RAN preRAN preRAN pre---designdesigndesign Static network simulation (design)Static network simulation (design)Static network simulation (design)

Design optimizationDesign optimizationDesign optimization Network optimizationNetwork optimizationNetwork optimization


4/65

4


imensioning

Dr.NicolasCotanis

September 20, 2004


UMTS: promises and challenges

Single traffic bearer CDMA technology is history

UMTS promises

> Countless services with assorted bandwidth, QoS and

traffic models> Increased capacity from MUD, smart antennas, ST

diversity, etc> Asynchronous operation

> Better multipath resolution for improving micro cell design RN design challenges

> A large set of bearers with selectable transmit formats(OVSF.etc)

> Traffic specification (one of the center pieces)> Design and optimization based on MC network simulation> New models for the new technologies


5/65

5


imensioning

Dr.NicolasCotanis

September 20, 2004


UMTS design strengths

Design strength

> Management of design inputs> Diversification of the UMTS bearers

for a given service (based onmobility and propagationenvironment)

> Good understanding of the arraysand statistics from simulation

> 2G CDMA design experience> UMTS test-bed involvement

Identify servicesTraffic engineering

Traffic models

Traffic environments

Create & distribute

terminals

RF configuration forNode-BsLocate Node-Bs

System parameters

Run MAPC

Check designperformance based

on arrays andstatistics

Map services to

UMTS bearers

A Business-plan and traffic review

B RAN design review

A

B

Design Input

RAN design

Traffic layer

RAN design

Simulation


6/65

6

Asystematicappro

achforUMTSRAND

imensioning

Dr.NicolasCotanis

September 20, 2004


UMTS design overview

Comprehend concepts before embarking for thedesign

List of services including- Mobility

- Propagation CH- Coverage type- Carrier

UMTS bearers

Operating point

(Eb/No) optimization

Terminals (singleservice per terminal)

Hardware

performance Many terminals perservice according tomobility, propagationCH, etc

Market Traffic environments- Number of terminals

- Morphology weighting

Traffic layer

Trafficenviron-ments.

- RF parameters per cell- Loading

Cell radius

Link Budget perservice type

Deploy sites

Static simulation- Number of snapshots

Generate arrays andreports

Design optimization

- RF parameter optimization- Include new sites, etc

Design Input

Traffic layer

RAN pre-design

Networksimulationandoptimization


7/65

7

Asystematicappro

achforUMTSRAND

imensioning

Dr.NicolasCotanis

September 20, 2004


IntroductionIntroductionIntroduction UMTS specific design inputs

Traffic layer

RAN preRAN preRAN pre---designdesigndesign Static network simulation (design)Static network simulation (design)Static network simulation (design)



8/65

8

Asystematicappro

achforUMTSRAND

imensioning

Dr.NicolasCotanis

September 20, 2004


Handling traffic

Terminal= a mobile unit that exercises one of the UMTSbearers for supporting a given service and abides to a setof hardware limitations (e.g. max TX power, Ec/Io, numberof fingers, etc)

Terminals are distributed for different services within themarket (traffic layer)

New concepts

Traffic classes

Traffic modeling

Orthogonality factors

Service operating point Power control errors

Traffic layer

RN design requires traffic models per subscriber


9/65

9

Asystematicappro

achforUMTSRANDimensioning

Dr.NicolasCotanis

September 20, 2004


Traffic classes

PS services are characterized by

> Bit rates, Average/Percentile delays, Subscriber traffic model,etc

Based on traffic model and delay, services are classified in four classes

> Conversational (C), Streaming (S)> Interactive (I), Background (B)

Not all QoS - functions implemented initially:

Real time (CS) connection for C and S classes Non real time scheduled (PS) for I and B classes

Traffic class Conversational Streaming Interactive Background

Fundamentalcharacteristics

Preserve time relationbetween informationentities of the streamConversational pattern(stringent and low delay)

Preserve timerelation betweeninformation entities ofthe stream

Request responsepatternPreserve payloadcontent (dataintegrity)

Destination is notexpecting the datawithin a certain timePreserve payloadcontent (dataintegrity)

Example of theapplication

Voice, video telephony,video games

Streaming media(audio and/or video)

Web browsing,network games

Backgrounddownload of e-mails


10/65

10

Asystematicappro


Dr.NicolasCotanis

September 20, 2004


Traffic modeling / problem statement

Traffic models must include

The data flow description during a session (including the request-response pattern of the end user),

The asymmetry The bandwidth (kilobits per second) requirement.

Traffic models for 2G networks were simple

(see www.lcc.com for Erlang-B calculator)

Poisson arrivals and exponential service time

mE per subscriber instead of bandwidth

No data flow description (only voice activity)

No asymmetry


11/65

11

Asystematicappro


Dr.NicolasCotanis

September 20, 2004


B-traffic, M/M/n models

M/M/n model for B-type traffic classes (e-mail, FTP, etc) Service class described by per user

> CH-rateR

> Average packet size E{l} (exponential)> Packet arrival rate o (Poisson)> Average delay objective E{T}

The minimum required number of channelsNfor the

objective E{T} is given by

> is the transmission efficiency due to ARQ,> n is the number of B-type traffic terminals> o = o/o is the subscriber link efficiency

( ) on

lETE

RN

+

+

1}{}{


12/65

12

Asystematicappro


Dr.NicolasCotanis

September 20, 2004


I-traffic, simulations

The multimedia traffic model is the basic model for I-traffic

Currently is not used by design tools for QoS evaluation

Probability of UL/DL active users, no. of terminals, etc

I-traffic parameters Sessions rate

Pages per session

Packets per page Reading time (RT)

Packet size

Usually, 3G tool has a single trafficmodel for all the PS type services.


13/65

13

Asystematicappro


Dr.NicolasCotanis

September 20, 2004


I-traffic simulation diagram

Statistical

Analysis

Traffic

generation

Packet

switching

Traffic load/sub User traffic profile

/ service

QoS

Reading

time

Session

Page

# CHs /

service


14/65

14

Asystematicappro


Dr.NicolasCotanis

September 20, 2004


I-traffic study case


15/65

15

Asystematicappro


Dr.NicolasCotanis

September 20, 2004


Orthogonality factor

3G: Path loss models + Propagation environments (channels)

Propagation channel := power delay profile (paths)

Propagation CH relevance

DL orthogonality factor ()

> =1 := perfectly orthogonal. No DL inter-codes interference

>


16/65

16

Asystematicappro


Dr.NicolasCotanis

September 20, 2004


Creating terminals

Service operating point

3G services may be transported over the air interface inmany different ways (bearers)

Transmit format(TF)

TF controls the bearers operating point (Eb/No @ BER or BLER)

The best bearer has the lowest operating point

Service mapping looks for the best bearer (Physical CH)to be used in each area of the market, or Eb/No for agiven bearer

RAN design requires operating points

They are not specified in the standards

May be obtained from vendors or by LLS

Bearer configuration Downlink

Service 144 kb/sFrame size 10 msInfo bits / frame 1440Bit per radio block * 120

Tail/CRC bits per radio block 8/0Turbo code Rate 1/3, 8 states

Decoding algorithm Max-Log MAPNumber of iterations 8Unequal repetition Not usedDTX 320 bitsOuter interleaving (10 ms) 7264 bitsDPCCH (pilot/TPC/TFCI) 16/8/8DPCCH-DPDCH power 0 dB

Spreading factor (DPDCH) 8Spreading factor (DPCCH) 8* A radio block is a group of bit to which a CRCword is appended.


17/65

17

Asystematicappro


Dr.NicolasCotanis

September 20, 2004


Creating terminals

Operating point variability

Less variability for voice bearersat low and medium speed

High variability for low mobilityand high data rate services

3

3.5

4

4.5

5

5.5

6

6.5

7

0 10 20 30 40 50 60

Mobile speed (km/h)

Eb/N

o(dB)@B

ER=1E-

VH-A

VH-B

PD-A

Bearer Environment Downlink Uplink

8kb/s kb/s Speech, BER = 10-

3km/hr, vehicular

500 km/hr, vehicular

8.2 dB,

15.6 dB

6.2 dB

13.6 dB64 kb/s LCD, BER = 10

-6120 km/hr, vehicular 5.3 dB 2.8 dB

144 kb/s LCD, BER = 10-6

120 km/hr, vehicular 5.8 dB 2.55 dB

384 kb/s LCD, BER = 10-6

Outdoor to indoor 3.5 dB 0.3 dB

64 kb/s UDD, BLER of 10% 120 km/hr, vehicular 5.2 dB Mainly uplink

144 kb/s UDD, BLER of 10% 120 km/hr 5.1 dB Mainly uplink

384 kb/s UDD, BLER of 10% Outdoor to indoor, 3km/hr 3.4 dB Mainly uplink

Cell capacity

M=15 @ 4.2 dB

M=8 @ 6.8 dB( )F

M+

=

1


18/65

18

Asystematicappro

achforUMTSRAN

Dimensioning

Dr.NicolasCotanis

September 20, 2004


Terminals

Creating terminals

The subscriber concept is replaced by the terminal concept

UMTS bearers after mapping services to transport channels

> Propagation CH (power delay profile)

> Mobility

Bearers operation point changes with propagation CH andmobility

Hint: one service per terminal

Hardware constrains (power class, body loss etc) map bearers intoterminals (according to the UMTS bearer used and power class, etc)

S e r v ic e X

U M T S B e a r e rO p t im u m = m in E b / N o

T e r m in a l

H a r d w a r e

c o n s t r a i n t s

M o b il i t yP r o p a g a t io n C H

T i l /U /A i


19/65

19

Asystematicappro

achforUMTSRAN

Dimensioning

Dr.NicolasCotanis

September 20, 2004


For each service

Subscribers (terminals)

Users (subscribers in session)

Active users (UL active = TX, DL active := RX)

CS services: P(A/U)= PS services: P(A/U)=Tcom/Tsession

Traffic engineering derives the size of the usersset

Per subscriber service load GoS or delays

Users are distributed within the market (usingclutter weighting or not)

Terminals/Users/Active users

Subscribersper service

Users

Active-users

Traffic load, QoS

Session profile

Trafficenineerin

InsideSimulato

r

P(A)=P(U) P(A/U)

Traffic layer


20/65

T ffi l


21/65

21

Asystematicappro

achforUMTSRAN

Dimensioning

Dr.NicolasCotanis

September 20, 2004


Traffic layer

Traffic layer

Provides the essential information for spreadingterminals. Traffic environments (TfE)

> business district, commercial district, city core, residential, transport

network,etc

Clutter weighting factors Number of terminals per TfE / route

Polygons are used for delimiting TfEor hot spots

Vectors are used for terminal distributionalong routes

TfE may be further divided according to

the propagation CH. type

High densityresidential

Commercialdistrict

Businessdistrict

Residential

GSM UMTS it h i


22/65

22

AsystematicapproachforUMTSRAN

Dimensioning

Dr.NicolasCotanis

September 20, 2004


Network configuration

GSM-UMTS site sharing

UMTS high data rate services may require between twoand three times as many base stations as do GSM.

Pros : environmentally friendly, quick and effective way to deploy

3G-

technology , cost cuts up to 40%. Cons : could amount to a single 3G-network, undermining competition

and technological innovation.

e.g. transceivers cannot be shared without sharing forecasting data

Forms of infrastructure sharing

Sharing of Node-B and RNC

Sharing of even the core network technology, including mobile switching

centers, IP routers, and location registries

Scenario RAN design

Headframe sharing (UMTS antennas) Site location and UMTS antenna height aregiven

Antennas sharing (dual-/tri-band antennas) As before plus UMTS antenna height,orientation and down tilt are given

Node-B equipment sharing As before plus number of channel elementsand service types to be coordinated withsharing partners


23/65

23


Dimensioning

Dr.NicolasCotanis

September 20, 2004


IntroductionIntroductionIntroduction UMTS specific design inputsUMTS specific design inputsUMTS specific design inputs

Traffic layerTraffic layerTraffic layer

RAN pre-design Static network simulation (design)Static network simulation (design)Static network simulation (design)


Key system parameters


24/65

24


Dimensioning

Dr.NicolasCotanis

September 20, 2004


Key system parameters

Key System Parameters Common Pilot Channel (CPICH) power

Synchronization Channel (SCH) powers

> Used by cell search (initial / target)

Cell loading factor SHO thresholds and gains

W-CDMA cells are identified by DL scrambling codes

W-CDMA cell search Code and time synchronization with the scrambling code of the best

server

Based on P-SCH, S-SCH* and CPICH

Powers for SCH and CPICH Tradeoff: cell capacity cell acquisition time

SCH/CPICH loading factors


25/65

25


Dimensioning

Dr.NicolasCotanis

September 20, 2004


SCH/CPICH loading factors

Cell search performance controlled by the power used(loading factors)

SCHSCHPSCHP PP =

( ) ( )hcCPiCHSCHSCHSCHIPPP ++=

( )hcCPICHSCHCPICHCPICH IPPP ++=

( ) ochcCPICHSCH IIPPG ++=

Geometry factor *

*Y-P Eric Wang, T. Ottoson Cell Search Algorithms and Optimization in W-CDMA, VTC-00, Spring 2000

For G=-3dB

5% < CPICH < 10%

SCH 10%

60% < P-SCH < 70%

CCH < 10%

IN unit

P_nodeB 43 dBm of P_nodeB 19.95 W

x_CPICH 10 % of P_nodeB

x_SCH 10 % of P_nodeB

x_PSCH 60 % of SCH

x_CCCH 10 % of P_nodeB

OUT

P_CPICH 33 dBm 2.00 W

P_SCH 33 dBm 2.00 W

P_PSCH 30.8 dBm 1.20 W

P_SSCH 29 dBm 0.80 W

P_CCCH 33 dBm 2.00 W

P_traffic 41.5 dBm 13.97 W

CPICH Power


26/65

26


Dimensioning

Dr.NicolasCotanis

September 20, 2004


Average acquisition time versus CPICH G= 3 dB,

P-SCH

=50%,

SCH=10%

5% < CPICH < 10%

Key System Parameters

CPICH Power

96.596.27108.13155.79

31.7431.932.1844.61

1462.5

33.2910

102

103

104

0 5 10 15

CPICH Loading

Averageacquisitiontime

(ms)

Initial

Target

SCH Power


27/65

27


Dimensioning

Dr.NicolasCotanis

September 20, 2004



SCH Power

Average acquisition time versus P-SCH G= 3 dB,

SCH=10%,

CPICH=10%

SCH 10%

60% < P-SCH < 70%

96.27

82.23

74.7469.63

67.18 68.41

31.9 31.6 31.5 31.6 31.8 32.320

30

40

50

60

70

80

90

100

110

0.45 0.55 0.65 0.75P-SCH Loading

Averageacquisition

time(ms)

Initial

Target

Cell Loading Factor


28/65

28


Dimensioning

Dr.NicolasCotanis

September 20, 2004



Cell Loading Factor

Noise rise (I) and loading factor () are related

Per propagation or traffic environment limits for loadingfactors

Link budget cell radius calculation

Static network simulation terminals disconnection

Hints

< 75% > 30% ; moderate cell breathing

DL> UL

Traffic environmentLoading factor

(%)Noise riseMAI (dB)

Rural and highways 30-40 1.55-2.23

Urban and dens-urban

75 6

=

+=

1

1log10log10dB

N

hcoc

P

III

Number of Rake fingers


29/65

29


Dimensioning

Dr.NicolasCotanis

September 20, 2004



Number of Rake fingers

The number of Rake fingers efficiency of utilizing

the incoming RF power (one finger per multipathcomponent) Requirement changes with the environment Fixed for a given terminal

The larger the number of Rake fingers in the terminal the better thesignal/noise

Measurements Typical urban (?): 4-6 fingers retrieve 80% of the Tx power for 95%

of the time. The rest of the Tx power (20%) becomes interference Dense urban (?): 10-15 fingers

No agreement on the required number of fingers andfinger assignment technique Single Rake finger assignment Grouped assignment

The number of fingers in the receiver must equal the size of the active set (dimensioning tool). In the absence ofmanufacturer data, the active set size should range between 4 to 6.In the case of grouped assignment, the active set must equal the number of groups and not the number ofindividual fingers

SHO thresholds and margins


30/65

30


Dimensioning

Dr.NicolasCotanis

September 20, 2004


UMTS handovers

SHO thresholds and margins

UMTS RAN designs require UL: S and So handoff margins (no thresholds) and gains DL: S/So handoff margins,gains and terminal AS size

UMTS uses relative SHO thresholds No need to tweak the parameters for different regions

Relative thresholds control SHO overhead Design tools use SHO marginsinstead of thresholds

The minimum difference between the CPICH signal of the bestserving sector and the pilot signal of all other sectors that are

considered for soft-handover communication SHO margin from SHO gain graphs.Functions of

> multipath profile (PL)

> terminal mobility and availability

of power control> BS/terminal antenna configuration

> receiver algorithm

ITU pedestrian A ITU vehicular ASingle radio link

Eb/No(dB)11.3 8.5

Macro diversityEb/No(dB)

7.3 / 8.6 6.3 / 7.7

Soft handover gain(dB) 4.0 / 2.7 2.2 / 0.8

Required Eb/No for FER=1%.CS 8 kbps bearer with full constant power

PL= 0 and 3 dB respectively

Link budget


31/65

31


Dimensioning

Dr.NicolasCotanis

September 20, 2004


Link budget

LB pre-dimensioning estimates the initial networkconfiguration

Based on many assumptions

Uniform inter-cell interference (f),

Soft handover overhead and gain Constant Node-B antenna gain Uniform traffic distribution and uniform service mapping Same results for every run, etc.

Link budgets are used for calculating the service radius foreach cell type and supported terminal

Based on the per services coverage objective (blanket orisland -where available-) cell type radii are determined

Typical Cell Ranges Outdoor, In-car @ 95% Indoor @ 80%

Bearer Urban Sub-urban Urban Sub-urban

12.2 kb/s Speech 1.98 km 3.34 km 1.02 km 2.55 km

64 kb/s 1.61 km 2.71 km 0.93 km 1.56 km

144 kb/s 1.42 km 2.39 km 0.62 km 1.04 km

384 kb/s 1.24 km 2.10 km 0.53 km 0.90 km

In the UMTS RAN design process; the linkbudget is used to estimate the calculation areafor each site/cell and not the actual cell radius.Thus, LB inaccuracies resulting from so manyassumptions are tolerable.

TerminalCoverage

radius (km)12.2 kbps speech 1.9864 kbps 1.61144 kbps 1.42Cell 1.42

Link Budget Model


32/65

32

Asystematicappr

oachforUMTSRAN

Dimensioning

Dr.NicolasCotanis

September 20, 2004


Link Budget

Link Budget Model

Nt

PTx,kLBODY EIRPTx,k

GTx

mLN

Lpen GRx

MAPLk

LRx

mTPC

SkgSS

(Eb/No)k

Icell

gSO

Matched

PRx,k

LTx

Only when thetransmitter is ahandheld

Nw

Receiverend

BearerCoverage

Transmitter end

Path loss model

Cell radius k

EIRP calculation

PRx,k calculation

Pk

NF NF


33/65

33

Asystematicappr

oachforUMTSRAN

Dimensioning

Dr.NicolasCotanis

September 20, 2004



Traffic layerTraffic layerTraffic layer RAN preRAN preRAN pre---designdesigndesign Static network simulation (design)


The SNS concept


34/65

34

Asystematicappr

oachforUMTSRAN

Dimensioning

Dr.NicolasCotanis

September 20, 2004


The SNS concept

W-CDMA RAN design tools based on the static network

simulation (SNS) concept. Snapshots of the RAN status at different instances in time Monte Carlo (MC) trials per snapshot for spatial distribution of

terminals*

Multiple access power control (MAPC) algorithms for assigning theamount of UL/DL power required by each radio link (RL)

> Connect or disconnect active terminals

SNS resolves assumptions

Estimates F-factor distribution SHO overhead and gain,etc

Uses antenna radiation pattern

Spatially distributes terminals

Output changes according to

terminal distribution

fF

hc

oc

I

I +=

+=

1

1

1

1

SNS output


35/65

35

Asystematicappr

oachforUMTSRAN

Dimensioning

Dr.NicolasCotanis

September 20, 2004


S S output

Arrays (plots): radionetwork attributesassociated to each bin

(e.g., CPICH coverage,Ec/It, effective servicecoverage, best server,SHO including active set,

etc

Statistics (databases), thestatus** of each Node-B orterminal in the working area

CPICH best server

SHO gain

SNS flowchart


36/65

36

Asystematicappr

oachforUMTSRAN

Dimensioning

Dr.NicolasCotanis

September 20, 2004


Static Network Simulation

Estimating the status of a CDMA radionetwork outside the history context

MAPC := an iterative algorithm that

finds out the best server and theactive set (AS) for each active terminaland, based on the type of service,provisions the minimum UL/DLtransmit power for supporting the radiolink

Randomly placeterminals and

assign services

Initial Best Server

(PL based)

For each active terminal doInitial UL/DL Power

Control

Best Server

(Ec/It based)

Convergence?

NoYes

Ready

For each active terminal do

UL/DL Power Control

The status of a CDMA radio network is represented by the setof terminals and services that are connected through the airinterface.

Randomize terminals


37/65

37

Asystematicappr

oachforUMTSRAN

Dimensioning

Dr.NicolasCotanis

September 20, 2004



Monte Carlo trials for distributingterminals within the working area

One service per terminal

pa:= Active (Tx or RX) state probability forservice*

nt := Number of k-type terminals in theworking area

na

:= Average number of active terminals


assign services

Initial Best Server

(PL based)


Control

Best Server(Ec/It based)

Convergence ?

NoYes

Ready


UL/DL Power Control

AtA pnn =

The number of terminals within a pixel follows a Poissondistribution with a mean value proportional with the

corresponding pixel area.

Path losses from a given Node-B to terminals within thesame bin may be different due to the randomnessinflicted by the shadowing

Initial Best Server and AS


38/65

38


Dimensioning

Dr.NicolasCotanis

September 20, 2004



The best server and the active set foreach terminal are initially calculatedbased on path loss (PL)

Initial Node-B noise


assign services

Initial Best Server

(PL based)


Control


Convergence?

NoYes

Ready


UL/DL Power Control

MAPC


39/65

39


Dimensioning

Dr.NicolasCotanis

September 20, 2004



For each terminal, MAPC finds the best server and

the AS, and assigns the minimum UL/DL transmitpower for providing the required service at thatinstance

Initial UL/DL power control for each active terminal

Initial Ec/It UL/DL Tx power per radio link based on

Terminal sensitivity including average power raise (PRx,k)

SHO gain

MAPC stops when convergence its reached orwhen exceeding a maximum number of iterations

Usable snapshot when MAPC converges

Different convergence criteria UL/DL criteria combined in a single indicator e.g. UL criterion


assign services

Initial Best Server

(PL based)


Control


Convergence ?NoYes

Ready


UL/DL Power Control


40/65

40


Dimensioning

Dr.NicolasCotanis

September 20, 2004



Does not disturb the system

> It does not contribute to the intercell and intracell interference> It does not consume channel resources

At the end of selected snapshots, the HT visits all the empty pixels(without terminals) for determining

> Best server

> Mean size of active set

> UL/DL required TX power, etc

> The most probable handover type

> Mean number of soft handover cells

The sector (site) related information required for UL/DL TX powercomputation (as the intracell and intercell interference, PA power etc.)are derived from statistics based on previous SNS snaphots.

Improper use of the hypothetical terminal may distort the SNS arrays and statistics (see Design Optimization).

The HT examines UL/DL service powers only. For example, the handover status display at a certain pixel mayindicate 3-way soft-handover however, a subscriber, which falls into this pixel, may or may not be in 3 wayhandover depending upon channel element availability

Options


41/65

41


Dimensioning

Dr.NicolasCotanis

September 20, 2004


SNS options copy the operation of a UMTS RAN

Carriers CPICH power allocation

Disconnecting terminals

If optional RF-CH are available

a) randomly assigning MSs within the entire networkb) randomly moving MSs from overloaded cells

c) moving only high transmit power MSs from overloaded cells

CPICH power allocation options

a) fix within the networkb) adjustable for each cell based on UL-interference level

c) cell selectable.

Exceeding Node-B total/RL transmit power

a) randomly disconnectb) disconnect the highest power links

c) disconnect the smallest power links


42/65

42


Dimensioning

Dr.NicolasCotanis

September 20, 2004



Traffic layerTraffic layerTraffic layer RAN preRAN preRAN pre---designdesigndesign Static network simulation (design)Static network simulation (design)Static network simulation (design)

Design optimization Network optimizationNetwork optimizationNetwork optimization

Questions after RAN design completion


43/65

43


Dimensioning

Dr.NicolasCotanis

September 20, 2004


Design optimization

Overall (network) performance Do the networks sites and sectors capture the projected

traffic? What is the throughput per Node-B? How many terminals of each service type are captured by

each sector in the network?

RF performance

What are the principal reasons for failure to connect? What is the percentage of RLs in SHO? What is the average downlink transmit power per service type

required?

Per service/Per carrier arrays


44/65

44


Dimensioning

Dr.NicolasCotanis

September 20, 2004


Design optimization

Per service arrays describe the serviceperformance Basically each bin denotes a probability

They may refer to many carriers Per carrier arrays describe the RF performance Are very similar with those used for 2G CDMA designs

Per service arrays


45/65

45


Dimensioning

Dr.NicolasCotanis

September 20, 2004


Design optimization

Coverage probability per service Best server by coverage probability

Coverage probability by service for the Nth best server

Path balance SHO arrays

Reason for failure by service

UL request TX power

Probability of CH limit failureDL/UL EbNoLow EcIoNo primary CH

Noise rise limitNo carrier

Service coverage probability


46/65

46


Dimensioning

Dr.NicolasCotanis

September 20, 2004


Design optimization

Application: shows the coverage probability forevery pixel in the simulation area(effective servicecoverage)

Input: -

Colors: Ranges for pS,B(%)

Description

Comments:

based solely on HTs in bins where ATs have not beenspread

BS

R

BS

CBS

N

NP

,

,, =

Service coverage probability (SCP)Best server by coverage probabilityCP for the Nth best server

Path BalanceReason for failure (by service)Probability of channel limit failureProbability of Downlink Eb/No failureProbability of Uplink Eb/No failureProbability of Low Ec/Io failureProbability of no carrier failureProbability of Noise Rise FailureUplink request Tx powerSoft handover arraysSecond order service arrays

Per carrier arrays (RF)


47/65

47


Dimensioning

Dr.NicolasCotanis

September 20, 2004


Design optimization / Carriers

Without simulation Best server by pilot Pilot strength

Pilot strength for the Nth best server After simulation Pilot Ec/Io Ec/Io for the Nth best server

Mean Io Mean received power Cell UL load

Pilot strengthPilot strength for the Nth best serverBest server by pilotPilot Ec/Io

Ec/Io for the Nth best serverAll serversCell UL loadMean IoMean received power


48/65

48

Asystematicapp

roachforUMTSRAN

Dimensioning

Dr.NicolasCotanis

September 20, 2004




Design optimizationDesign optimizationDesign optimization Network optimization

UMTS network optimization


49/65

49

Asystematicapp

roachforUMTSRAN

Dimensioning

Dr.NicolasCotanis

September 20, 2004


Measurements required for UMTS networkoptimization Delay profile (usually before design) Cell search time Frame error rate and radio propagation MS transmission power DHO

Uplink capacity

Delay profiles


50/65

50

Asystematicapp

roachforUMTSRANDimensioning

Dr.NicolasCotanis

September 20, 2004


Wide band channelmeasurements

The link quality decreaseswhen the number ofmultipath are larger thenthe number of fingers

Optimization: Antenna down tilt or better

sectorization

Cell search time


51/65

51

Asystematicapp


Dr.NicolasCotanis

September 20, 2004


MSs always monitorneighboring CPICH duringcommunication for thepurpose of DHO control.

MSs have to complete thisprocess as fast as possible.

The ration between CPICH

power and MAI controls thecell search time

Frame error rate and radio propagation


52/65

52

Asystematicapp


Dr.NicolasCotanis

September 20, 2004


Measurements ofRSSI, SIR, FER andtotal available pathnumber

MS transmission power


53/65

53

Asystematicapp


Dr.NicolasCotanis

September 20, 2004


UMTS supports SIRbased fast TPC tosolve the near/farproblem.

MSs transmissionpower gives anindication of the TPCefficiency on reducingMAI and increasingnetwork capacity

Optimization: TPC step size, rate, etc

DHO


54/65

54

Asystematicapp


Dr.NicolasCotanis

September 20, 2004


The number ofconnectioncells/sectors in theservice area are

measured

Optimization: Thadd and Thdel parameters

Uplink capacity


55/65

55

Asystematicapp


Dr.NicolasCotanis

September 20, 2004


Measurements on theuplink interferencelevel

The lower the MAI thehigher the uplinkUMTS capacity

Optimization: MUD, smart antennas, etc


56/65

56

Asystematicapp


Dr.NicolasCotanis

September 20, 2004




Design optimizationDesign optimizationDesign optimization Network optimizationNetwork optimizationNetwork optimization UMTS overlay

Objectives


57/65

57

Asystematicapp


Dr.NicolasCotanis

September 20, 2004


Design optimization

UMTS overlay = deploying UMTS on top of an existentGSM network securing all the benefits of a synergeticnetwork.

Overlay design = Constrains on UMTS design

Benefits of overlay

Reducing deployment coststhrough co-location/co-sitting

Controlling the NB interference(avoiding near-far effects)

Resource sharing; coverageextension, capacity sharing,service distribution

Issues to address

UMTS network evolution, Coverage, Co-location, Interference and noise, Inter-system handover, Mobility management,

Traffic load sharing, etc

Launch architecture

I t t d CN GERAN


58/65

58

Asystematicapp


Dr.NicolasCotanis

September 20, 2004


Integrated CN

upgrades and uses the MSC androuting elements of the legacyGSM/GPRS network

Careful deployed for preserving the

capacity and performance of theGSM/GPRS revenue-earningnetwork

BTS

BSS

BTS

RNC

Node B

UTRAN

Node B

SGSN

GPRSVLR

MSC

VLR

Gb

A

Iu-ps

Iu-cs

2G and 3G

CS-elements

2G and 3G

PS-elements

BTS

BSS

BTS

GERAN

RNC

Node B

UTRAN

Node B

SGSN

GPRSVLR

MSC

VLR

Gb

A

SGSN

GPRSVLR

MSC

VLR

Iu-ps

Iu-cs

2G CS-elements

2G PS-elements

3G CS-elements

3G PS-elements

Segregated CN uses a new network of switching

and routing elements to supportUMTS

UMTS overlay deployment

T j h


59/65

59

Asystematicapp


Dr.NicolasCotanis

September 20, 2004


Two major phases

radio coverage

service availability.

Inspect technical/real estate issues for co-location

Multi-band antennas?

WCDMA site and antenna allocationInter-system interference analysisLink budget analysis per service

GSM-WCDMA traffic load distribution

Combined GSM/WCDMA analysis

WCDMA design

Objective reached

Review:site configuration,

traffic,design objectives

Far from objective

Design performance?

Review GSM network coverage

Select candidate sites (Spectrum carving if required)

Coverage analysis

Utili ti f i ti 2G it ( l ti ) d h ki f


60/65

60

Asystematicapp

roachforUMTSRA

NDimensioning

Dr.NicolasCotanis

September 20, 2004


Utilization of existing 2G-sites (co-location) needs checking for

Holes in the 2G-coverage (for basic services as voice)

Increased intra-system interference for WCDMA should be avoided..

Unit

GSM900/

speech

GSM1800/

speech

WCDMA/

speech

WCDMA/

144 kbps

WCDMA/

384 kbps

Mobile Tx power dBm 33 30 21 21 21

Receiver sensitivity 1 dBm -110 -110 -124 -117 -113

Interference margin 2 dB 1 0 2 2 2

Fast fading margin 3 dB 2 2 2 2 2

BS antenna gain 4 dBi 16 18 18 18 18

Body loss 5 dB 3 3 3

MS antenna gain 6 dBi 0 0 0 2 2Relative path loss gainwith frequency 7

dB 11 1

MAPL dB 164 154 156 154 150

1 GSM sensitivity includes receive antenna diversity. WCDMA

sensitivity is calculated based on the equation

dBdB

EbNoSFkTBF + )(log1010

, where bandwidth

B=3.84 MHz, spreading gain SF=B/(data rate), and F=4dB is

the WCDMA base station noise figure.

2 An interference margin of 1 dB was used for GSM because of

the small amout of spectrum in GSM 900 that does not allow

large reuse factors. For a loading of=37%, the noise raise is

dB2)1(log10 10 =

3 The reduced fast fading margin comes from including the macro

diversity gain

4 Three sector configuration are assumed for both GSM andWCDMA

5 Data terminals have not to stay close to the users had

6 Antenna gain for data terminals is 2 dBi

7 Represent variations in the path loss attenuation with frequency

versus the UMTS Region 1 band

Intersystem interference

WCDMA i t f


61/65

61

Asystematicapp

roachforUMTSRA

NDimensioning

Dr.NicolasCotanis

September 20, 2004


WCDMA interference

Interference from 2G systems

Mobile RxMobile Tx

Mobile RxMobile Tx

Mobile RxMobile Tx

900880 920 940 960 980

960935915890

GSM

PCNEU17501700 1800 1850 1900 1950

1710 1785 1805 1880

18001750 1850 1900 1950 2000

1850 1910 1930 1990

PCSUS

Frequency (MHz)

Frequency (MHz)

Frequency (MHz)

Unit Region 1 Region 2

Up-link

(MS transmit,base receive)

MHz 1920 - 1980 1850 1910

Down-link(MS receive,base transmit)

MHz 2110 2170 1930 - 1990

2G interference and background noise

Interference mechanisms from adjacent 2G bands


62/65

62

Asystematicapp

roachforUMTSRA

NDimensioning

Dr.NicolasCotanis

September 20, 2004


Interference mechanisms from adjacent 2G bands

ACI for up-links

ACI, intermodulation, and wide band noise in the down-link

WCDMA-2G coordination

Background noise measurements Identifying co-location issues (as

imperfect spectrum clearance orleakage from adjacent bandwidth)

Estimating the sensitivity reductiondue to background noise rise fromheavy motor traffic(~ 1.9 dB in urban environments)

2G WCDMA 2G2G 2G

Operator 3 Operator 1 Operator 2

2G WCDMA2G 2G


WCDMA2G 2G


WCDMA

A

B

C

Reduction in coverage and capacity

Coverage and capacity reductions investigated through


63/65

63

Asystematicapp

roachforUMTSRA

NDimensioning

Dr.NicolasCotanis

September 20, 2004


Coverage and capacity reductions investigated through

Monte Carlo simulations A denser WCDMA network will be more resistant to 2G interference

Guard bands of approximately 1MHz should eliminate 2G interferenceissues (applicable for non-coordinated 2G systems)

If the legacy 2G-network is made of micro cells and the WCDMA cells arelarge ( 1.5 km) guard bands cannot alleviate capacity reductions.

WCDMA-2G collocation has a doubleadvantage: it reduces deploymentcosts and builds coordination forminimizing 2G Interference risks

Conclusions


64/65

64

AsystematicapproachforUMTSRA

NDimensioning

Dr.NicolasCotanis

September 20, 2004


3G designs rely on preliminary steps

Overlay designs require

Extra preliminary steps Design tools for investigating GSM-UMTS synergy

3G Design

Tool

Traffic dataw/ QoS control

Configuration

Key NetworkParameters

Servicemapping

SNS

LLS

3G-Traffic

WB

sounding

Thank you for attending!


65/65

65

AsystematicapproachforUMTSRA

NDimensioning

Dr.NicolasCotanis

September 20, 2004


Presentation-IIR04-Tutorial-A Systematic Approach for Wcdma Design

Documents