Huawei RNC Capacity dimension

Capacity dimension
Principle for capacity dimension
Input from operator ( Total subscribers in network, traffic model) RNC capacity limitation ( Erl, Throughput,BHCA)

Number of WRBS

Radio Network dimension

If Nodebs per WRBS >100

Subscribers per WRBS

Subscribers per Nodeb

Nodebs per WRBS

Traffic Model Parameters

Traffic Parameter BHCA CS per sub BHCA total per sub CS Voice per sub/BH CS Data per sub/BH CS Video Telephony call duration/BH CS call duration/BH Rel.99 PS per sub/BH Total PS per sub/BH Timer for CH switching (DCH/FACH) Location update rate sub/BH Ratio of soft/softer HO/BH Proportion of HO/BH Handover times/call/BH (inter/intra RNC SHO) PS CH switching Cell update/PSCall Ratio of UL : DL (Rel.99/HSDPA) 0.6 0.9 16,2mErl 1,1mErl 135sec 135sec 580bps 1160bps 600-800msec 2.2 0.8:0.2 30% 7,0 times/call 4.0 times/call 3 times/call 1:3.6 Value

Traffic Model from Vodafone

3

Limitation
Limiting factor : throughput capacity 60Mbps/ WRBS

e.g.

120Mbps / 60Mbps = 2

2 WRBS

Limiting factor : BHCA capacity

75,000/ WRBS

e.g.

60,000 / 75,000 = 0.8

1 WRBS

Limiting factor : NodeB capacity 150 / 100 = 1.5

100 NodeB/ WRBS 2 WRBS

e.g.

Limiting factor : Cell capacity

300 NodeB/ WRBS

e.g.

400 / 300 = 1.3

2 WRBS

RNC Dimension
2*63 channalized E1s

Max. Capacity for different RNC types Traffic capacity Interfaces RNC types Configure Iub Mbps 77 206 461 NodeB 100 300 600 Cell 300 900 1800 STM-1 E1(ATM) 30 26 20 64 192 384 E1(IP) 64 192 384 FE 32 96 192 Channelize d STM-1 2 6 12

Small Capacity RNC 1WRSS+1WRBS Medium Capacity RNC 1WRSS+3WRBS High Capacity RNC 1WRSS+6WRBS

Traffic Capacity
Iu CS Iu 12.2 Iu CS Data Mbit/s AMR Erl Erl Iu PS I/B 64/128 Mbit/s 28.50 86.07

RNC Types Configure Subscribers BHCA Small Capaticy RNC 1 WRSS+ 1 WRBS 95000 85500 50.19 923.4 52.25 Media Capacity RNC 1 WRSS+ 3 WRBS 287000 258300 151.56 2789.64 157.85 High Capacity RNC 1 WRSS+ 6 WRBS 574000

Iur Mbit/s 7.654 20.623

516600 303.10 5579.28 315.7

172.13

46.162

VF CALL PROFILE DATAS

RNC Capacity Limitation

CS Erlang and CS Throughput
There define a relationship between CS Erlang and throughput as follows: 1 CS Erlang = 16kbps throughput. The relation comes from the processing consuming in WFMRb board.

CS Throughput and PS Throughput

For each RNC subrack :
CS throughput + PS throughput <= 60Mbps; CS throughput < = 40Mbps;

RNC Capacity Limitation

BHCA calculation from CS 40Mbps
Supposing each call duration is 120s. The maximum CS throughput is 40Mbps, equals to 2500Erl.

Erl = BHCA * Call duration /3600

It can derive that : BHCA = 2500/(120/3600) = 75,000 Equivalent BHCA 1. This index is used to measure the signaling processing capability in WSPUb. 2. The equivalent BHCAconsiders the procedure not only the call procedure, but also other signalling procedure, including SMS, IMSI attachment, location update, handover, cell update, URA update, etc. 3. The equivalent BHCA comes from the test data where the WSPUb CPU load is up to 80%. 4. The maximum equivalent BHCA is 360k and the WSPUb CPU load is 80%.

Note: The actual signaling capacity is limited by Integrated BHCA

7

RNC Capacity Limitation

Equivalent BHCA
Basic Signaling CS BHCA per BM per BH PS BHCA per RNC per BH Location Area Update per RNC per BH Routing Area Update per RNC per BH Soft Handover Addition per RNC (intra/inter RNC) Soft Handover Deletion per RNC (intra/inter RNC) Softer Handover Addition per RNC (intra/inter RNC) Softer Handover Deletion per RNC (intra/inter RNC) CS Inter-RAT handover per RNC PS channel switching per RNC SMS per RNC per BH Paging per RNC per BH RRC REJ Equivalent BHCA 1.00 1.00 0.46 0.46 0.26 0.13 0.18 0.09 0.56 0.41 0.46 0.01 0.15

RNC Capacity Limitation

Maximum Subscribers number
The maximum number of subscriber depends on the Integrated BHCA, call model and maximum data throughput:
Procedure multiplied per Cost 360k

equivalent BHCA(# Subscriber, call model) < Maximum Equivalent BHCA

60M per subRack

Total throughput ( #Subscriber, call model) < Maximum data throughput

Board information
WFMRb
1. 2. WFMRb board is user plane processing board, handling the user data of AAL2 cell for Iu/Iub/Iur interface. One RNC subrack contains 10 WFMRb with the maximum capability of 60Mbps for PS throughput or 40Mbps for CS throughput

3.
4. 5.

The maximum user plane capability of WFMRb board in RAN5.1 is 6Mbps for PS packet processing. (MAC layer 7.2M)
For CS voice processing, 4Mbps can be handle for each WFMRb board. WFMRb board contributes to the RNC total PS throughput as well as the CS throughput.

10

Board information
WSPUb
1. 2. WSPUb board is signalling processing board which handle the RANAP/ RNSAP/SS7/QAAL2/RRC/NBAP signaling; Each WSPUb board could handle 75,000 BHCA (for typical traffic model) which does not includes the location update, paging, handover, SMS, attaching/detaching, etc. procedure. The maximum equivalent BHCA is 360k and the WSPUb CPU load is 80%.

3.

WOSEc
1.
2.

WOSEc board is Iub interface board.

Each WOSEc board can process 60Mbps throughput.

11

Board Information
Summary
1. 2. WFMRb and WSPUb are the main boards contributing to RNC capacity; WOSEc board is the Iub interface board which is not the limitation for the RNC capacity;

3.

WMUXb board is the switching board in each sub-rack and is not the limitation for the RNC capacity;

12

Some analysis for New Years eve

1.

Between 00:00-00:15, Altogether 21000 CS call attempts processed by RNC 121, but only 13000 RAB assigned by CN, so somewhere in CN must be overload.

2.

Between 00:00-00:59, Altogether 51055 RRC connect rejected, among them 49683 belong to SMS, only 1372 belong to voice call.

Conclusion: Between 00:00-00:15, both RNC and CN are overloaded. RNC is overloaded (RRC rejection) with only 600 Erlangs because abnormal high number of SMS and Core network lack of reply.

13

RRC EVOLUTION NEW YEAR CHARTS

RRC.FailConnEstab.Cong 60000 50000 40000 30000 20000 10000 0
2006-12-31 21:00:00 2006-12-31 22:00:00 2006-12-31 23:00:00 2007-01-01 00:00:00 2007-01-01 01:00:00 2007-01-01 02:00:00 2007-01-01 03:00:00 2007-01-01 04:00:00 2007-01-01 05:00:00 2007-01-01 06:00:00 2007-01-01 07:00:00 2007-01-01 08:00:00 2007-01-01 09:00:00 2007-01-01 10:00:00

SMS
160000 140000 120000 100000 80000 60000 40000 20000 0

RRC Attemps New year

locUpdate

CS call orig

SumaDeRRCAttConnEstabUnknow n SumaDeRRCAttConnEstabTmStrCall SumaDeRRCAttConnEstabTmLw PrSig SumaDeRRCAttConnEstabTmInterCall SumaDeRRCAttConnEstabTmHhPrSig SumaDeRRCAttConnEstabTmConvCall SumaDeRRCAttConnEstabTmBkgCall SumaDeRRCAttConnEstabReg SumaDeRRCAttConnEstabOrgStrCall SumaDeRRCAttConnEstabOrgInterCall SumaDeRRCAttConnEstabOrgConvCall SumaDeRRCAttConnEstabOrgBkgCall SumaDeRRCAttConnEstabOgSubCall

interRAT

2006-12-31 20:00:00

2006-12-31 23:00:00

2007-01-01 02:00:00

2007-01-01 05:00:00

2007-01-01 08:00:00

2007-01-01 11:00:00

2007-01-01 14:00:00

2007-01-01 17:00:00

2007-01-01 20:00:00

2007-01-01 23:00:00

2007-01-02 02:00:00

2007-01-02 05:00:00

2007-01-02 08:00:00

2007-01-02 11:00:00

2007-01-02 14:00:00

2007-01-02 17:00:00

SumaDeRRCAttConnEstabOgLw PrSig SumaDeRRCAttConnEstabOgHhPrSig SumaDeRRCAttConnEstabIRATCelRes SumaDeRRCAttConnEstabIRATCCO SumaDeRRCAttConnEstabEmgCall SumaDeRRCAttConnEstabDetach SumaDeRRCAttConnEstabCallReEst

14

Mapping nodeb to SubRack

40 nodeb on SubRack1 Half of NodeB on Cpu1

15

SPU load evolution

CPU WSPU1:1 100% 80% 60% 40% 20%

30/12/2006 12:00 30/12/2006 16:00 30/12/2006 20:00 31/12/2006 13:00 31/12/2006 17:00 31/12/2006 21:00

VS.MeanCPUUtil.SPU VS.MaxCPUUtil.SPU

01/01/2007 10:00

01/01/2007 14:00

01/01/2007 18:00

01/01/2007 22:00

30/12/2006 0:00

30/12/2006 4:00

30/12/2006 8:00

31/12/2006 1:00

31/12/2006 5:00

31/12/2006 9:00

01/01/2007 2:00

01/01/2007 6:00

0%

WFMRName WFMR:1:1 WFMR:1:12 WFMR:1:13 WFMR:1:14 WFMR:1:2 WFMR:1:3 WFMR:1:4 WFMR:1:5 WFMR:1:6 WFMR:1:9

Time(As hour) VS.MeanCPUUtil.FMR VS.MaxCPUUtil.F 01/01/2007 0:00 26% 31% 01/01/2007 0:00 29% 34% 01/01/2007 0:00 26% 30% 01/01/2007 0:00 26% 31% 01/01/2007 0:00 27% 33% 01/01/2007 0:00 26% 31% 01/01/2007 0:00 29% 34% 01/01/2007 0:00 24% 29% 01/01/2007 0:00 27% 33% 01/01/2007 0:00 25% 30%

More space on User plane than Signalling plane

16

SPU Dimension using current CPU load

Total nber of subcribers in network: 1000k 600Erl Erlangs_RNC121/TotalErlangs=2% =>20k Subs in RNC121 Erlangs for each Nodeb from counters Nodeb to SPU map( worse SPU 40 nodeb among 87) The cpu load is 8% for common task if the traffic load is 0 Cpu load with 275 Erlangs=40% at BH and limit is 80% => the SPU can handle 275*(80-8)/(40-8)=618 Erlangs => If 618/87=7 Erlangs per nodeB => a SPU can handle 550/7=78Nodeb

Erlangs per SPU =600*40/87=275 Erl

17