Update From PMWG

8/11/2019 Update From PMWG

http://slidepdf.com/reader/full/update-from-pmwg 1/49

Company Confidential

1 © 2005 Nokia Update from BSS PMWG / November 2005 / Poul Larsen

Update from BSS Performance

Management Working Group (PMWG)

OSP Seminar, Bangkok, November 9 –

10, 2005

Poul Larsen

OSP, Expertise Leverage





Contents

• Planning/Benchmark KPI set for S10.5 & S11

• How to measure throughput with Nokia BSS counters• Current issues with UL EDGE RLC block counters

• Current issues with multislot request/allocation counters

• S11.5 counters & KPIs





Planning/Benchmark KPI set for S10.5 & S11

In order to standardize the use of GPRS/EDGE BSS counters in Nokia networks, two KPIsets have been defined by BSS PMWG:

• Planning KPI set

• Should be used by the network planning/performance department to optimize the BSS part ofthe network

• Currently contains 53 KPIs

• Network Doctor report 290 (ND for OSS4)

• Benchmark KPI set

• Should be used to compare networks

• Currently contains 18 KPIs (subset of planning KPI set)

• Published as part of system documentation

• Network Doctor report 802 (ND for OSS4)

• Updated KPI lists available via Jump:http://nea.ntc.nokia.com/dld.php/?BSS_Benchmark_KPIs_BSSPMWG.xls

http://nea.ntc.nokia.com/dld.php/?BSS_Planning_KPIs_for_PS_BSSPMWG.xls

• OSP guideline (including S11.5 KPIs) available via QuickPlace:• http://qp2.connecting.nokia.com/QuickPlace/npcommunityqp/PageLibraryC2256FB8004C916E.nsf/h_9538C43F6B64CACBC2256FB80076F160/98E3C6BEF84A2A8BC22570A400489156/?OpenDocument





How to measure throughput

with Nokia BSS counters





How to measure throughput with Nokia BSS counters

• Throughput based on RLC block counters and TBF duration

• Throughput based on RLC block counters• Throughput based on MS flowrate

• Volume-weighted LLC throughput





Formulas based on RLC blocks and TBF durationE.g. trf_72b:

(RLC_data_blocks_UL_CS1*20 + RLC_data_blocks_UL_CS2*30)* 8/1000

-----------------------------------------------------------------------------------------

ave_dur_UL_TBF_sum/100

---------------------------------------------------------------------------------------------------------------------------

alloc_1_TSL_UL + 2*alloc_2_TSL_UL + 3*alloc_3_TSL_UL + 4*alloc_4_TSL_UL)

-----------------------------------------------------------------------------------------------

alloc_1_TSL_UL + alloc_2_TSL_UL + alloc_3_TSL_UL + alloc_4_TSL_UL

• ave_dur_UL_TBF_sum does not include abnormally released TBFs=> Makes the formula show too high values (even 100s of kbps)

• ave_dur_UL_TBF_sum includes the TBF release delay where no data is transmitted

=> Makes the formula show too low values

• Original plans to modify ave_dur_xx_TBF counters in S11.5 have been discarded

• ave_dur_UL_TBF_sum contains both GPRS and EDGE TBFs

=> Formula just for GPRS-only networks• Denominator is updated at TBF establishments and reallocation. Duration is not taken into account





Throughput based on RLC blocks (I)

E.g. trf_72d (GPRS UL effective throughput pr timeslot):

(RLC_data_blocks_UL_CS1*20 + RLC_data_blocks_UL_CS2*30)*8 /1000

---------------------------------------------------------------------------------------------------------------------------------------------------------------

(rlc_data_blocks_ul_cs1 + rlc_data_blocks_ul_cs2 + rlc_mac_cntrl_blocks_ul + BAD_FRAME_IND_UL_CS1 +BAD_FRAME_IND_UL_CS2

+ BAD_FRAME_IND_UL_UNACK + IGNOR_RLC_DATA_BL_UL_DUE_BSN) /50

• RLC/MAC control blocks from downloading applications produce low UL values• Ignored blocks happen at the end of a TBF. Not noticed by user, but pulls the

formula values down

• Since “Bad frames” counters are used, e.g. many cell reselections will make thisformula give low values

• Typical values in GPRS networks:

• UL: 2 - 4 kbps

• DL: 11 – 12 kbps





Throughput based on RLC blocks (I)

(UL: tr f_72d & DL: tr f_73d)

• The ratio between data blocks and control blocks important for the value of the

formulas• Short TBFs will have relatively more control blocks than long TBFs

• Upper-layer data transfer will typically generate long TBFs in one direction and

many RLC acknowledgements in the opposite direction

• E.g. for MMS:

10 MMS sent from cell A to cell B (in perfect radio conditions)

MO BTS MT BTS

UL CS2 12946 508

UL Ignored 727 1379

UL Ctrl 528 1231

DL CS2 332 10493

DL Ctrl 972 653

UL throughput (trf_72d) 10.8 2.0

DL throughput (trf_73d) 3.1 11.3





Throughput based on RLC blocks (II)

• Replace bad frames with retransmissions

• Remove effect from applications (remove ignored and control blocks)=> trf_233b

(rlc_data_blocks_ul_cs1 - rlc_data_blocks_ul_unack)*20 + rlc_data_blocks_ul_cs2*30)*8 /1000

----------------------------------------------------------------------------------------------------------------------------------------------------------------------------

rlc_data_blocks_ul_cs1 - r lc_data_blocks_ul_unack + rlc_data_blocks_ul_cs2 + retra_data_blocks_ul_cs1 + retra_data_blocks_ul_cs2) /50

• Cell reselection and congestion not included

• KPI useful for determining the quality of the link (radio + abis)





UL GPRS RLC throughput (trf_233b)

((rlc_data_blocks_ul_cs1 - rlc_data_blocks_ul_unack)*20 + rlc_data_blocks_ul_cs2*30)*8 /1000

----------------------------------------------------------------------------------------------------------------------------------------------------------------------------

rlc_data_blocks_ul_cs1 - r lc_data_blocks_ul_unack + rlc_data_blocks_ul_cs2 + retra_data_blocks_ul_cs1 + retra_data_blocks_ul_cs2) /50

FTP Upload

0

2

4

6

8

10

12

14

~50 10 5 3

Uplink C/I (dB)

k b i t / s / t i m e s l o t

FTP Tp

RLC Tp

• Tests from Nokia Test Network, April 2005

• C/I values only approximate





DL GPRS RLC throughput (trf_235a)

((rlc_data_blocks_dl_cs1 - rlc_data_blocks_dl_unack)*20 + rlc_data_blocks_dl_cs2*30)*8 /1000

----------------------------------------------------------------------------------------------------------------------------------------------------------------------------

rlc_data_blocks_dl_cs1 - r lc_data_blocks_dl_unack + rlc_data_blocks_dl_cs2 + retra_data_blocks_dl_cs1 + retra_data_blocks_dl_cs2) /50

• Tests from Nokia Test Network, April 2005

• C/I values only approximate

FTP Download

0

2

4

6

8

10

12

14

~50 10 5

Downlink C/I (dB)

k b i t / s / t i m e s l o t

FTP Tp

RLC Tp





Downlink EDGE RLC throughput (trf_236)

• Unit is kbit/s (pr timeslot)• trf_236 shows low values when low MCS are used and/or many

retransmissions take place.

(sum over MCS - 1 (xx)*22+ + sum over MCS - 2 (xx)*28+

+ sum over MCS - 3 (xx)*37+ + sum over MCS - 4 (xx)*44+ + sum over MCS - 5 (xx)*56+ + s um over MCS - 6 (xx)*74+ + sum over MCS - 7 (xx)*56+ + sum over MCS - 8 (xx)*68+ + sum over MCS - 9 (xx)*74)*8/1000 --------------------------------------

sum over msc1…6 of (yy)/50 + sum over msc7…9 of (yy)/2/50

wherexx = dl_rlc_blocks_in_ack_mode

yy = dl_rlc_blocks_in_ack_mode + retrans_rlc_data_blocks_dl





FTP download with EDGE

•Taken from lab tests in Nokia Test Network

• Nokia 6220 (2 + 1)

• Good UL quality

• MCA = 7

DL Link Quality KPIs

0%

20%

40%

60%

80%

100%

~50 17 14 11 7

Downlink C/I

MCS9

MCS8

MCS7

MCS6

MCS5

MCS4

MCS3

MCS2

MCS1

Throughput pr timeslot

0

10

20

30

40

50

60

70

~50 17 14 11 7

Downlink C/I (dB)

k b i t / s

FTP Throughput(kbit/s)

RLC Throughput(kbit/s)





HTTP download with EDGE

• Large file (890 KB, multiple objects) simulates web browsing• Small file (1 KB, text only) simulates WAP browsing

• NTN, N6220, good UL quality

• End-user experience largely independent of link quality

• RLC throughput gives good view of link quality

• Small files (= short TBFs) doesn’t give much time for linkadaptation to adjust (MCA = 7)

• Remember that also EDAP & GPRS/EDGE multiplexing willdowngrade coding scheme

HTTP, Large Files

0%

20%

40%

60%

80%

100%

~50 20 15 10 5

Downlink C/I

MCS9

MCS8

MCS7MCS6

MCS5

MCS4

MCS3

MCS2

MCS1

HTTP, Small Files

0%

20%

40%

60%

80%

100%

~50 20 15 10 5

Downlink C/I

MCS9

MCS8

MCS7

MCS6

MCS5

MCS4

MCS3

MCS2

MCS1

Throughput pr timeslot

0

10

20

30

40

50

60

~50 20 15 10 5

Downlink C/I (dB)

k b i t / s

HTTP Tp, Largefiles

RLC Tp, Largefiles

HTTP Tp, Smallfiles

RLC Tp, Smallfiles





Live network statistics

Network 1 Network 2 Network 3

Downlink GPRS RLC throughput (trf_235a) 11.8 kbit/s 11.9 kbit/s 11.9 kbit/s

Uplink GPRS RLC throughput (trf_233a) 12.0 kbit/s 12.0 kbit/s 12.0 kbit/s

Downlink EDGE RLC throughput (trf_236) 43 kbit/s 29 kbit/s 34 kbit/s

Uplink EDGE RLC throughput (trf_234) 32 kbit/s 14 kbit/s 17 kbit/s

GPRS RLC Throughput

10.00

10.50

11.00

11.50

12.00

10.00 10.50 11.00 11.50 12.00

Downlink

U p l i n k

Uplink





Live network statistics

Network 1 Network 2 Network 3

Downlink GPRS RLC throughput (trf_235a) 11.8 kbit/s 11.9 kbit/s 11.9 kbit/s

Uplink GPRS RLC throughput (trf_233a) 12.0 kbit/s 12.0 kbit/s 12.0 kbit/s

Downlink EDGE RLC throughput (trf_236) 43 kbit/s 29 kbit/s 34 kbit/s

Uplink EDGE RLC throughput (trf_234) 32 kbit/s 14 kbit/s 17 kbit/s

Downlink EGPRS RLC throughput (kbits/s/tsl) TRF_236A

0,00

10,00

20,00

30,00

40,00

50,00

60,00

2 0 0 5 / 0 6 / 2 7 0 0

2 0 0 5 / 0 6 / 2 7 1 9

2 0 0 5 / 0 6 / 2 8 1 4

2 0 0 5 / 0 6 / 2 9 0 9

2 0 0 5 / 0 6 / 3 0 0 4

2 0 0 5 / 0 6 / 3 0 2 3

2 0 0 5 / 0 7 / 0 1 1 8

2 0 0 5 / 0 7 / 0 2 1 3

2 0 0 5 / 0 7 / 0 3 0 8

2 0 0 5 / 0 7 / 0 4 0 3

2 0 0 5 / 0 7 / 0 4 2 2

2 0 0 5 / 0 7 / 0 5 1 7

2 0 0 5 / 0 7 / 0 6 1 2

2 0 0 5 / 0 7 / 0 7 0 7

2 0 0 5 / 0 7 / 0 8 0 2

2 0 0 5 / 0 7 / 0 8 2 1

2 0 0 5 / 0 7 / 0 9 1 6

2 0 0 5 / 0 7 / 1 0 1 1

2 0 0 5 / 0 7 / 1 1 0 6

2 0 0 5 / 0 7 / 1 2 0 1

2 0 0 5 / 0 7 / 1 2 2 0

2 0 0 5 / 0 7 / 1 3 1 5

2 0 0 5 / 0 7 / 1 4 1 0

2 0 0 5 / 0 7 / 1 5 0 5

2 0 0 5 / 0 7 / 1 6 0 0

2 0 0 5 / 0 7 / 1 6 1 9

2 0 0 5 / 0 7 / 1 7 1 4

2 0 0 5 / 0 7 / 1 8 0 9

2 0 0 5 / 0 7 / 1 9 0 4

2 0 0 5 / 0 7 / 1 9 2 3

2 0 0 5 / 0 7 / 2 0 1 8

2 0 0 5 / 0 7 / 2 1 1 3

2 0 0 5 / 0 7 / 2 2 0 8

2 0 0 5 / 0 7 / 2 3 0 3

2 0 0 5 / 0 7 / 2 3 2 2

2 0 0 5 / 0 7 / 2 4 1 7

BSC1 BSC2 BSC3

Uplink EGPRS RLC throughput (kbits/s/tsl) TRF_234A

0,00

5,00

10,00

15,00

20,00

25,00

30,00

35,00

40,00

2 0 0 5 / 0 6 / 2 7 0

0

2 0 0 5 / 0 6 / 2 7 1

9

2 0 0 5 / 0 6 / 2 8 1

4

2 0 0 5 / 0 6 / 2 9 0

9

2 0 0 5 / 0 6 / 3 0 0

4

2 0 0 5 / 0 6 / 3 0 2

3

2 0 0 5 / 0 7 / 0 1 1

8

2 0 0 5 / 0 7 / 0 2 1

3

2 0 0 5 / 0 7 / 0 3 0

8

2 0 0 5 / 0 7 / 0 4 0

3

2 0 0 5 / 0 7 / 0 4 2

2

2 0 0 5 / 0 7 / 0 5 1

7

2 0 0 5 / 0 7 / 0 6 1

2

2 0 0 5 / 0 7 / 0 7 0

7

2 0 0 5 / 0 7 / 0 8 0

2

2 0 0 5 / 0 7 / 0 8 2

1

2 0 0 5 / 0 7 / 0 9 1

6

2 0 0 5 / 0 7 / 1 0 1

1

2 0 0 5 / 0 7 / 1 1 0

6

2 0 0 5 / 0 7 / 1 2 0

1

2 0 0 5 / 0 7 / 1 2 2

0

2 0 0 5 / 0 7 / 1 3 1

5

2 0 0 5 / 0 7 / 1 4 1

0

2 0 0 5 / 0 7 / 1 5 0

5

2 0 0 5 / 0 7 / 1 6 0

0

2 0 0 5 / 0 7 / 1 6 1

9

2 0 0 5 / 0 7 / 1 7 1

4

2 0 0 5 / 0 7 / 1 8 0

9

2 0 0 5 / 0 7 / 1 9 0

4

2 0 0 5 / 0 7 / 1 9 2

3

2 0 0 5 / 0 7 / 2 0 1

8

2 0 0 5 / 0 7 / 2 1 1

3

2 0 0 5 / 0 7 / 2 2 0

8

2 0 0 5 / 0 7 / 2 3 0

3

2 0 0 5 / 0 7 / 2 3 2

2

2 0 0 5 / 0 7 / 2 4 1

7

BSC1 BSC2 BSC3





Conclusion on RLC throughput formulas

• Not end-user experience

• Application independent• GPRS formulas suitable for determining link quality (radio & abis)

• EDGE formulas suitable for determining link quality (radio & abis) if:

• MCA same value in whole network

• TBF length (ie application mix) similar for all cells

• No EDAP capacity problems (actually also valid for GPRS!)

• No multiplexing with GPRS handsets





BSSGP flow rate

MS specific flowrate (trf_125, ND report 270):

8/1000 * ave_ms_bssgp_flow_rate_sum

---------------------------------------------------

ave_ms_bssgp_flow_rate_den

• Is calculated for each QoS class (Best Effort, Silver, Gold)

• Only available for downlink

• Does not take timeslot allocation and technology (GPRS/EDGE) into account

• Counts only when there is a TBF => Does not take the effect of hard

blocking/preemption & cell reselections into account• Good correlation with application throughput, but

• Includes the idle periods during the transfer => e.g. WAP browsing produces smallvalues => Typical values from live networks are low (a few kbps) => Operator needsto use QoS classes to get useful values

• Currently a couple of issues with counters:

• UL QoS class not mapped correctly when using GPRS

• QoS class OK when MS used as modem, but not OK when application is on MS (fixedend-October?)





BSSGP Flowrate

• Test results from NTN lab

• EDGE 3+1 handset• HTTP browsing

• Good correlation between

trf_125 and application

throughput• For larger files, good

correlation between radio

link conditions and

throughput

• File size has big impact onthroughput values

~50 dB 10 dB 5 dB

1 KB, App 4.0 3.5 3.0

1 KB, trf_125 6.0 6.2 5.1

10 KB, App 21.3 15.4 9.3

10 KB, trf_125 22.6 18.2 13.550 KB, App 33.6 19.7 11.7

50 KB, trf_125 37.2 24.8 17.4

100 KB, App 33.8 19.9 12.6

100 KB, trf_125 39.5 26.5 18.1

1000 KB, App 112.5 35.1 18.5

1000 KB, trf_125 123.3 45.1 23.8

HTTP throughput vs trf_125

0

20

40

60

80

100

120

140

~50 dB 10 dB 5 dB

Downlink C/I

k b i t / s

1 KB, App

1 KB, trf_125

10 KB, App

10 KB, trf_125

50 KB, App

50 KB, trf_125

100 KB, App

100 KB, trf_125

1000 KB, App

1000 KB, t rf_125





Volume Weighted LLC Throughput - Counter Principle• In S11, two new counters are incremented only for EDGE mobiles having 4 TSL

multislot capability in DL

• Also when MS gets less than 4 TSLs in DL, because of small GPRS territory or concurrentUL&DL allocation

• Specific requirements are needed for Bytes and Duration counters:

• Short TBFs not taken into account

• based on byte and duration threshold (>4 sec. OR >1560 bytes)

• Unit of duration of 10 ms

• Abnormally released TBFs are NOT considered -> Better Accuracy

• Throughput calculation principle

• VW LLC throughput = Sum of (Payloads per traffic burst * Throughput per traffic burst) /Payload

• Traffic burst (burst)= Continuous data transfer from data arrives to PCU buffer till PCUbuffer gets empty

• New counters

• spare72109 = Sum of (Payloads per burst * Throughput per burst) of EDGE 4 TSL MS (Units : 100B* B/10ms)

• spare72110 = Payload of EDGE 4 TSL MS (Unit: 100B)• spare counters are used to minimize impacts in BSC - NetAct interface

• Counters ready from S11, CD6.0





Volume weighting - example

Burst A: Payload = 100 Kbyte, duration = 8 seconds => throughput = 100 kbit/s

Burst B: Payload = 10 KByte, duration = 16 seconds => throughput = 5 kbit/s

Average throughput:

Payload / duration = (800 Kbit + 80 Kbit) / (8 seconds + 16 seconds) = 37 kbit/s

Volume Weighted throughput:

((800 Kbit * 100 kbit/s) + (80 Kbit * 5 kbit/s)) / (800 Kbit + 80 Kbit) = 91 kbit/s

Throughput for long TBFs are weighted more than for short TBFs





VW LLC Throughput Formula

• Counters

• spare72109 =Sum of (Payloads per burst * Throughput per burst) of EDGE 4 TSL MS ( Units :100Byte* Byte/10ms)

• spare72110 = Payload of EDGE 4 TSL MS (Unit: 100Byte)

72109 (100 Byte * Byte/10ms) * 8 (bits/Byte)*(100 units of 10ms/second)

• LLC Throughput (kbits/s) = ----------------------------------------------------------------------- =

72110 (100Byte) * 1000(bits/kbit)

spare72109 * 8

= ------------------------------ kbits/s = llc_3

spare72110*10





VW LLC Throughput TC1- FTP DL

Case 1 FTP Download (Good) Case 1 FTP Download (Bad)

• The graphs below show the VW DL LLC throughput for FTP DL performed with

different file sizes and under different radio conditions

• As file size decreases application throughput is reduced due to TCP slow start

effect.

• Volume Weighted LLC Throughput follows the application throughput, but the

impact of TCP slow start is practically eliminated by the VW LLC throughput

counters

0

50

100

150

200

250

F T P D L

2 . 5

M B ( 1 . 1

)

F T P D L 1 M B

( 1 . 2

)

F T P D L

5 0 0 K B ( 1 . 3

)

F T P D L

1 0 0 K B ( 1 . 4

)

k b i t s / s Weighted LLC Thr

Application Thr

0

50

100

150

200

250

F T P D L

2 . 5

M B

F T P D L

1 M B

F T P D L

5 0 0 K B

F T P D L

1 0 0 K B

k b i t s / s Weighted LLC Thr

Application Thr





Performance of VW LLC Throughput

• The graph below plots the VW LLC throughput under different radio conditions

and with different traffic mixes

• VW LLC Throughput keeps quite steady independently on traffic mix

• Only if FTP uploads were highly dominant (TC2), VW LLC Throughput would be

impacted by the traffic mix

0

50

100

150200

250

TC 1.2 FTP

DL 1MB

TC 3.1 RF

FTP 1MB

DL + 1MB

UL

TC 3.2 RF

Mixed

traffic

TC 3.3 RF

Mixed

traffic

WEB

Traffic

ONLY

Good

Bad

VW LLC Throughput





VW LLC throughput at BSC level (Busy Hour)

0

20

40

60

80

100

120

140

160

180

200

220240

0 8 .

1 1 .

2 0 0 5

0 8 .

1 2 .

2 0 0 5

0 8 .

1 3 .

2 0 0 5

0 8 .

1 4 .

2 0 0 5

0 8 .

1 5 .

2 0 0 5

0 8 .

1 6 .

2 0 0 5

0 8 .

1 7 .

2 0 0 5

0 8 .

1 8 .

2 0 0 5

0 8 .

1 9 .

2 0 0 5

0 8 .

2 0 .

2 0 0 5

0 8 .

2 1 .

2 0 0 5

0 8 .

2 2 .

2 0 0 5

0 8 .

2 3 .

2 0 0 5

0 8 .

2 4 .

2 0 0 5

0 8 .

2 5 .

2 0 0 5

14

17

Average of VW LLC Throughput (Daily)

BSC_NAME Date

Hour

•Problems in low traffic BTSs





VW LLC throughput at BTS level (Hourly)

0

50100

150

200

250

300

350

400

450

500

550

600

650

700

750

800

0 10000 20000 30000 40000 50000 60000 70000

LLC Payload (KB)

T h r o u g h p u

t ( k b p s )

VW LLC Throughput (Daily)

•5.6% of the samples show values >236kbps (402 from 7146 non 0samples)

•Problems in low-traffic BTSs

•Problem identified, fix is being tested





Volume Weighted LLC throughput

• Counters can be used to show the potential throughput that a 4 timeslot EDGE

MS would get, if it is downloading big files

• Low KPI values can be caused by many things

• Bad radio link quality

• Not enough timeslot allocation

• Too much timeslot sharing

• Bad transmission link quality• Not enough EDAP capacity

• Etc.

• Note: Cell reselection & hard blocking do not have impact

=> Further analysis needed on bad performing areas

V l W i ht d LLC th h t





Volume Weighted LLC throughput

• 4 tsl EDGE DL counters available from S11, CD6.0 (all QoS classes)

• In S11.5, CD2.3, VW LLC throughput counters will be available for:• 4 tsl EDGE MS

• Other-than-4 tsl EDGE MS

• GPRS MS

• Counters available for each QoS class

• Counters available for UL and DL





Current issues with UL EDGE

RLC block counters

P ti t i i





Preemptive retransmissions in GPRS:

Preemptive retransmissions

Block X Block Y Block Z Block Z Block Z Block Z


Transmitted by MS

Received by PCU

RLC_DATA_BLOCKS_UL_CSx IGNOR_RLC_DATA_BL_UL_DUE_BSN

P ti t i i





Preemptive retransmissions in EDGE:

Preemptive retransmissions



Transmitted by MS

Received by PCU

UL_RLC_BLOCKS_IN_ACK/UNACK_MODE IGNOR_RLC_DATA_BL_UL_DUE_BSN &

UL_RLC_BLOCKS_IN_ACK/UNACK_MODE

• No longer possible to accurately measure UL Payload

• Problem present since S10.5ED, but impact only understood recently

• By making certain assumptions, new formulas can be defined that tries to

compensate for this effect

Impact on formulas





Impact on formulas

Example - UL EDGE RLC payload (trf_214a):

sum over MCS-1 (xx)* 22+sum over MCS-2 (xx)* 28+sum over MCS-3 (xx)* 37+

sum over MCS-4 (xx)* 44+sum over MCS-5 (xx)* 56+sum over MCS-6 (xx)* 74+

sum over MCS-7 (xx/2)* 112+sum over MCS-8 (xx/2)* 136+sum over MCS-9 (xx/2)* 148

------------------------------

1024

where xx = (ul_rlc_blocks_in_ack_mode + ul_rlc_blocks_in_unack_mode)

Impact on formulas





Impact on formulasModified version - UL EDGE RLC payload (trf_214b):

sum over MCS-1 (xx)* 22+






sum over MCS-7 (xx/2)* 112+

sum over MCS-8 (xx/2)* 136+

sum over MCS-9 (xx/2)* 148

------------------------------ 1024

where

xx = (ul_rlc_blocks_in_ack_mode +

ul_rlc_blocks_in_unack_mode –

“ignor_rlc_data_bl_ul_due_bsn_mcs_n”)

and where

“ignor_rlc_data_bl_ul_due_bsn_mcs_n” =“ignor_rlc_data_bl_ul_due_bsn_egprs” *

(sum over MCSn (UL_RLC_Blocks_In_Ack_Mode +

UL_RLC_Blocks_In_Unack_Mode)/

sum over MCS1-9 (UL_RLC_Blocks_In_Ack_Mode +

UL_RLC_Blocks_In_Unack_Mode))

and where

“ignor_rlc_data_bl_ul_due_bsn_egprs” = trf_204a ignor_rlc_data_bl_ul_due_bsn * ----------------

trf_203+ trf_204a

and where

trf_203 = rlc_data_blocks_ul_cs1 + rlc_data_blocks_ul_cs2

and

trf_204a =

-(RLC_DATA_BLOCKS_UL_CS1 + RLC_DATA_BLOCKS_UL_CS2)

+ sum over MCS1 to 9 (UL_RLC_BLOCKS_IN_ACK_MODE + UL_RLC_BLOCKS_IN_UNACK_MODE)

- IGNOR_RLC_DATA_BL_UL_DUE_BSN

+ sqrt (

(RLC_DATA_BLOCKS_UL_CS1 + RLC_DATA_BLOCKS_UL_CS2

- sum over MCS1 to 9 (UL_RLC_BLOCKS_IN_ACK_MODE + UL_RLC_BLOCKS_IN_UNACK_MODE)

+ IGNOR_RLC_DATA_BL_UL_DUE_BSN) *

(RLC_DATA_BLOCKS_UL_CS1 + RLC_DATA_BLOCKS_UL_CS2

- sum over MCS1 to 9 (UL_RLC_BLOCKS_IN_ACK_MODE + UL_RLC_BLOCKS_IN_UNACK_MODE)

+ IGNOR_RLC_DATA_BL_UL_DUE_BSN)

+ 4 * sum over MCS1 to 9 (UL_RLC_BLOCKS_IN_ACK_MODE +

UL_RLC_BLOCKS_IN_UNACK_MODE)

* (RLC_DATA_BLOCKS_UL_CS1 + RLC_DATA_BLOCKS_UL_CS2))

-------------------------------------------------------------------------------

2

Impact of current counter implementation





Impact of current counter implementation

• Modified formulas become very complicated – difficult to implement in tools and difficult to

explain to operators

• Even modified formulas only provide estimates

• Probably very drastic impact on KPI values

• Affected formulas in BM and planning KPI set have been modified (but many other

affected formulas not modified)

• Currently it is being discussed to change counter triggering in CD to S11.5

KPI Old

formulas

Modified

formulas

UL EDGE RLC payload, Kbyte (trf_214a & trf_214b) 1358145 133898

UL EDGE Erlang (trf_161d & trf_161g) 1526 357

UL GPRS Erlang (trf_205 & trf_205c) 1679 2691

PS Erlang (trf_237 & trf_237c) 3205 3048

UL EDGE RLC throughput, kbit/s/timeslot (trf_234 & trf_234a) 19.7 16.9





Current issues with multislot

request/allocation counters

Multislot request/allocation counters






• Before S11, CD6.0 & S11.5, CD2.0, the counters were updated in this way:

• Example: Class 6 MS can be in either 3+1 or 2+2 configuration – but never 3+2

• DL is “preferred” direction - MS gets a 3+1 allocation

• Counters updated as follows:

• REQ_2_TSL_UL

• REQ_3_TSL_DL

• ALLOC_1_TSL_UL

• ALLOC_3_TSL_DL

• In this case, counters show false blocking







• From S11, CD6.0 & S11.5, CD2.0, onwards, the counters are updated in this way:

• Example: Class 6 MS can be in either 3+1 or 2+2 configuration – but never 3+2

• DL is “preferred” direction - MS gets a 3+1 allocation• Counters updated as follows:

• REQ_1_TSL_UL

• REQ_3_TSL_DL

• ALLOC_1_TSL_UL

• ALLOC_3_TSL_DL

• In this case, counters do not show false blocking

• But: If territory was small, MS could also have gotten a 2+2 allocation:

• Counters updated as follows:

•REQ_1_TSL_UL

•REQ_3_TSL_DL

•ALLOC_2_TSL_UL

•

ALLOC_2_TSL_DL• In this case, counters show negative UL blocking





S11.5 counters & KPIs

S11 5 counters & KPIs





S11.5 counters & KPIs

• Comprehensive list of new counters, KPIs and Network Doctor reports in Ari

Saarinen’s presentation at PM Seminar, Espoo, September 2005:

• Following slides show only selected counters/KPIs

• New S11.5 counters require OSS4 to be stored in NetAck database

• Network Doctor reports available for the most important KPIS

• S11.5 KPIs not yet part of recommended KPI list due to lack of live network

experience

PS Erlangs





PS Erlangs

• Traditional way to count PS Erlangs: Count number of blocks and divide with timeslotcapacity (50 blocks pr second):

• New counters use alternative way of calculating PS Erlangs

• 72168 AVE_BUSY_GPRS_CH_UL

• 72169 AVE_BUSY_GPRS_CH_DL

• 72170 AVE_BUSY_GPRS_CH_DEN

• KPIs

trf_238 = AVE_BUSY_GPRS_CH_UL / AVE_BUSY_GPRS_CH_DEN

trf_239 = AVE_BUSY_GPRS_CH_DL / AVE_BUSY_GPRS_CH_DEN

sum over msc1…6 of (dl_rlc_ blocks_in_ack_mode +dl_rlc_blocks_in_unack_mode + retrans_rlc_data_blocks_dl)

+ sum over msc7…9 of (dl_rlc_blocks_in_ack_mode +dl_rlc_blocks_in_unack_mode + retrans_rlc_data_blocks_dl)/2+ “rlc_mac_cntrl_blocks_dl_egprs”-------------------------------- ---------------------------

period_duration*60*50

sum over msc1…6 of (dl_rlc_ blocks_in_ack_mode +dl_rlc_blocks_in_unack_mode + retrans_rlc_data_blocks_dl)

+ sum over msc7…9 of (dl_rlc_blocks_in_ack_mode +dl_rlc_blocks_in_unack_mode + retrans_rlc_data_blocks_dl)/2+ “rlc_mac_cntrl_blocks_dl_egprs”-------------------------------- ---------------------------

period_duration*60*50

EGPRS TBF in GPRS territory & vice versa





EGPRS TBF in GPRS territory & vice versa

• In case of separate EDGE and GPRS territories within same segment, itbecomes interesting to know if each territory is correctly dimensioned

•72117 UL_GPRS_TBF_IN_EGPRS_TERR

• 72118 DL_GPRS_TBF_IN_EGPRS_TERR

• 72119 UL_GPRS_TBF_FOR_EGPRS_REQ

• 72120 DL_GPRS_TBF_FOR_EGPRS_REQ

• EDGE resources used for GPRS TBFs:

tbf_57 = UL_GPRS_TBF_IN_EGPRS_TERR / (UPLINK_TBF - EGPRS_TBFS_UL)

tbf_58 = DL_GPRS_TBF_IN_EGPRS_TERR / (DOWNLINK_TBF - EGPRS_TBFS_DL)

• EDGE TBFs has been allocated GPRS resources:

tbf_59 = UL_GPRS_TBF_FOR_EGPRS_REQ / (REQ_1_TSL_UL_FOR_EGPRS_MS + REQ_2_TSL_UL_FOR_EGPRS_MS +

REQ_3_TSL_UL_FOR_EGPRS_MS + REQ_4_TSL_UL_FOR_EGPRS_MS + REQ_5_TSL_UL_FOR_EGPRS_MS + REQ_6_TSL_UL_FOR_EGPRS_MS +REQ_7_TSL_UL_FOR_EGPRS_MS + REQ_8_TSL_UL_FOR_EGPRS_MS)

tbf_60 = DL_GPRS_TBF_FOR_EGPRS_REQ / (REQ_1_TSL_DL_FOR_EGPRS_MS + REQ_2_TSL_DL_FOR_EGPRS_MS +REQ_3_TSL_DL_FOR_EGPRS_MS + REQ_4_TSL_DL_FOR_EGPRS_MS + REQ_5_TSL_DL_FOR_EGPRS_MS + REQ_6_TSL_DL_FOR_EGPRS_MS +REQ_7_TSL_DL_FOR_EGPRS_MS + REQ_8_TSL_DL_FOR_EGPRS_MS)

TBFs for signaling





TBFs for signaling

• Especially in new networks, a potential large part of the traffic will consist of

signaling traffic (GMM & SM)

• To tune parameters like routing area borders and SGSN security procedures, it is

of interest to operators to know the amount of RLC payload caused by signaling

• With the new counters we can find how many TBFs are used by signaling

• 72161 UL_TBF_FOR_SIGNALLING

• 72162 UL_TBF_FOR_DATA

•72163 DL_TBF_FOR_SIGNALLING

• 72164 DL_TBF_FOR_DATA

• KPIs

tbf_6 = UL_TBF_FOR_SIGNALLING / (UL_TBF_FOR_SIGNALLING + UL_TBF_FOR_DATA)

tbf_6 = DL_TBF_FOR_SIGNALLING / (DL_TBF_FOR_SIGNALLING + DL_TBF_FOR_DATA)

DL 8PSK to GMSK





• In case an uplink GPRS TBF uses same timeslots as a downlink EDGE TBF, the

EDGE TBF is forced to use GMSK modulation – which normally gives degraded

performance compared to 8PSK modulation• 72167 DL_8PSK_TO_GMSK_DUE_UL_GPRS

rlc_6 = DL_8PSK_TO_GMSK_DUE_UL_GPRS /

sum over msc1...9 of

( dl_rlc_blocks_in_ack_mode + dl_rlc_blocks_in_unack_mode )

MCS limited by EDAP





y

• Previously, the “EDAP congestion” counters included both EDAP and PCUlimitations. Now it is possible to separate the two issues

• 76019 UL MCS LIMITED BY PCU• 76020 DL MCS LIMITED BY PCU

dap_7a =sum( a.dl_tbfs_with_inadeq_edap_res -a.DL_MCS_LIMITED_BY_PCU) / (50 * 60)

-------------------------------------------------

sum over BTS with EGENA = Y

( rlc_data_blocks_dl_cs1 *20

+ rlc_data_blocks_dl_cs2 *30+ sum over MCS-1 (xx)* 22

+ sum over MCS-2 (xx)* 28





+ sum over MCS-7 (xx/2)*112



+ sum over MCS-11(xx)*36 + sum over MCS-12(xx)*50 ) / (1024 * 1024 * 1024)

where xx = (dl_rlc_blocks_in_ack_mode + dl_rlc_blocks_in_unack_mode)

MCS limited by PCU





y

• Previously, the “EDAP congestion” counters included both EDAP and PCUlimitations. Now it is possible to separate the two issues

dap_9 =sum( a.DL_MCS_LIMITED_BY_PCU) / (50 * 60)

-------------------------------------------------

sum over BTS with EGENA = Y


+ rlc_data_blocks_dl_cs2 *30










+ sum over MCS-11(xx)*36 + sum over MCS-12(xx)*50 ) / 1024

where xx = (dl_rlc_blocks_in_ack_mode + dl_rlc_blocks_in_unack_mode)

dap_10

sum( a.UL_MCS_LIMITED_BY_PCU) / (50 * 60)

-------------------------------------------------sum over BTS with EGENA = Y


...

S11.5 features Network Controlled Cell Reselection

(NCCR) and Network Assisted Cell Change (NACC)





(NCCR) and Network Assisted Cell Change (NACC)

• PCU1 and PCU2

• Reports and KPIs

• 282 GPRS cell reselection

• nnn RX level and quality report (planned, not implemented yet)

• several new KPIs, see following slides

• Measurements & counters• Cell reselection (95)

• object source BTS + target cell

• GPRS RX Level and Quality (96)

• object TRX + coding scheme

• Packet Control Unit• 6 counters

GPRS Cell Reselection Measurement





Object level:

Started per criteria

Power budget-

GPRS MS 95002

Service based

ISNCCR 95004

Coverage based

ISNCCR 95005

Quality Control 95006

Failures

Successful

Flush from

SGSN for NC2

95012

PCCO timer

expiry 95013

NCCR succ rel to

old cell 95014

Trigger fulfilled,but no cellchange ordersent. Targetcell id =candidate

Attempts(=NCCRtriggerFulfilled)

Not started due AC

95015

Optional NACC usage

With NC0 95017

Flush for NC0 =

Flush msgs

received (72206)

– flush for NC2

(95011)

TBF notreleased

No resp from

target cell 95007

Assignment

reject 95008

Ongoing CS

connection

95009

GMM standby or

forced 95010

Other 95011 Power budget-

EGPRS MS 95003

QC trig no good

neighbor 72204

Serv ISNCCR nogood neighbor

72205

PCU measurement, targetnot available

Setup failures

Total attempts =

sum(setup

failures)+sum(start

ed per criteria)

QC attempts =

95005 + 72204

Serv ISNCCR

attempts = 95003

+ 72205

Setup phase

Ave duration sum

95016

With NC2 95018

Total Flush

messages 72206

Source BTS-

id

Target LAC-

CI PLMN id 95000

QC = Quality Control

global RNC id

95001

• Global RNC is relevant only, if

target is 3G cell

p_nbsc_cell_reselection

Reportedby MS

noticedby PCU

GPRS RX Level & Quality Measurement

ND report not

available yet,

http://nea.ntc.nokia.com/cgi-bin/cgiwrap/neva/counter_list_bsc.pl?counter_id=95002
















































Uplink results with the same principle

•Measurement object is BTS + TRX + Coding Scheme

•RX level to RX level class conversion: adjustable RX level boundaries

•Measurement is based on RX level and quality measurements done by MS and BTS

•BEP to RX quality class conversion (fixed mapping) for EGPRS MSs

•NCCR feature must be active, DL RX level information received in Enhmeasurement reports (only) in NC2 mode

•This measurement generates lots of data!

RX Quality Class 0 RX Quality class 1 ETC RX Quality class 7

RX Level CS-1 CS-2 CS-3… MCS-9CS-1 CS-2 CS-3… MCS-9 CS-1 CS-2 CS-3 … MCS-9

0

12

3

4

5

0…10 11…20 21...30

RX1 RX2

31...40 41...50 Above 51

RX3 RX4 RX5

Class 0 Class 1 Class 2 Class 3 Class 4 Class 5

nr of samples

downlink

y ,

schedule:

12/05?





Thanks!

Questions?

[email protected]

Documents

Update From PMWG