GPRS/EGPRS RN Optimization Workshop

PM Detailed (R10 & R11)

Martin Skarve
Senior Specialist
Ericsson Services, Stockholm

EAB-05:001190 Uen Rev A 2005-01-17 1

IP Throughput
KPI versus PI Level 1

Av. # TS Av. utilization TBFs per Radio Link

reservable for all TBFs PDCH Bitrate
per TBF

MAXTSDL MUTILBASIC DLTBFPBPDCH CS12DLACK

TRAFF2BTBFSCAN X TRAFF2BTBFSCAN X DLBPDCH X CS12DLSCHED * 0,02

=
IP Throughput
DLBGGTHR
DLBGGDATA

EAB-05:001190 Uen Rev A 2005-01-17 2

 LIVE

Well?
CELL IP_THR_D_G IP_DISCARD IP_VOL_D_G RLBR_DL_B TBF_PDCH_D MUT_MAX MUT_BASIC PDCH_FAIL PREEMP_PCH PDCHS PDCHS ACT
T04MM8+ 12,93 10,53 9292,00 11,85 1,20 3,03 84,70 0,65 0,00 3,46 1,09
T04MM7+ 13,13 6,32 14829,00 11,54 1,19 3,09 90,07 1,13 0,00 4,70 1,88
T49M3+ 13,30 3,86 8108,00 28,70 1,26 2,70 66,99 46,27 320,00 1,57 0,57
T04MM9+ 14,75 6,28 26818,00 11,76 1,28 3,06 92,84 16,02 15,00 6,44 3,19
T80M3+ 15,77 6,24 10642,00 11,40 1,25 3,31 94,75 8,50 11,00 3,62 1,25
T86M2+ 16,73 8,03 11206,00 11,79 1,25 2,64 86,34 32,03 62,00 3,59 1,33
T51M7+ 18,57 10,64 9917,00 11,74 1,19 1,90 94,19 0,17 0,00 4,25 1,67
TMZZ3+ 19,09 3,25 30261,00 12,02 1,44 3,34 84,86 36,50 443,00 4,60 2,15
T80M9+ 22,02 7,35 9048,00 11,63 1,16 2,45 88,58 6,29 0,00 2,90 0,89
T04MM6+ 22,51 5,62 21539,00 11,73 1,19 3,09 90,29 1,48 0,00 4,94 1,92
T80M7+ 22,89 5,03 14809,00 11,69 1,17 2,87 93,32 4,20 0,00 3,98 1,57
T51MB+ 23,57 5,23 11409,00 11,85 1,10 2,86 77,71 53,85 143,00 2,29 1,05
TMZZ9+ 24,13 4,84 32173,00 11,90 1,31 2,45 83,09 44,06 249,00 4,29 1,80
T04MK9+ 24,40 7,55 11082,00 12,06 1,14 1,86 89,71 12,13 0,00 2,47 0,74
TMZZ7+ 24,58 6,93 14530,00 11,74 1,20 2,17 92,82 10,35 25,00 4,36 1,71
3,09 TS x 90,07%TS utilization / 1,19 sharing x 11,54 kbit/s & TS
=>
27 kbit/s = 13,13 kbit/s

2,17 TS x 92,82%TS utilization / 1,20 sharing x 11,74 kbit/s & TS

=>
19 kbit/s = 24,58 kbit/s
EAB-05:001190 Uen Rev A 2005-01-17 3
IP Throughput – in PCU IP Throughput – in STS

t2 - t1 includes TBF set-up time and

acknowledgement time at TBF end.
LLC-packets = x kbit

Throughput
t1 PCU in STS

Large TBF =>

Normal throughput
in STS
Whole TBF – ack. t2 sec
~2-300ms TBF
Um size
x
= Throughput kbit/sec
t2 - t1 Small TBF =>
Low throughput in STS Break-point
~2 sec? TBF

EAB-05:001190 Uen Rev A 2005-01-17 4

IP Throughput DL – GPRS & EGPRS
Reference
IP Throughput DLBG(G,E)DATA
DLBG(G,E)THR
DLBG(G,E)DATA DLBG(G,E)DATA
DLBG(G,E)PFC

Av. # TS Av. utilization TBFs per Radio Link

reservable for all TBFs PDCH Bitrate
per TBF

MAXTSDL MUTIL(BASIC,GPRS,EGPRS)UL DLTBFP(B,G,E)PDCH Σ(xx * INTxxBR(E)GPRSTBF)

TRAFF2x(B,G,E)TBFSCAN TRAFF2(B,G,E)TBFSCAN DL(B,G,E)PDCH Σ(INTxxBR(E)GPRSTBF)allxx

CS12DLACK
Update in R11 CS12DLSCHED * 0,02
Other formulas apply when QoS etc. is implemented

EAB-05:001190 Uen Rev A 2005-01-17 5

Multislot counters
MAXTSDL
TRAFF2x(B,G,E)TBFSCAN Av. # TS reservable per TBF

MUTIL(BASIC,GPRS,EGPRS)
TRAFF2(B,G,E)TBFSCAN Av. Utilization for all TBFs

Steps during both active and non-active TBFs (R10)

No TBF upgrade during non-active TBF
Only active TBFs counted in R11

Includes GMM/SM (from R10 AC-A2 only 1 TS is allocated for GMM/SM)

It is possible to exclude this in formula by using MUTILxy counters
( (MUTIL12 + MUTIL22) * 2 + (MUTIL13 + MUTIL23 + MUTIL33 ) *3 + ( MUTIL14 + MUTIL24 + MUTIL34 + MUTIL44 ) * 4)
MUTIL12 + MUTIL22 + MUTIL13 + MUTIL23 + MUTIL33 + MUTIL14 + MUTIL24 + MUTIL34 + MUTIL44

Small transfers (<500B) are excluded from R11

Includes QoS Streaming
Streaming, EIT and DTM excluded from R11
EAB-05:001190 Uen Rev A 2005-01-17 6
Sharing counters
DLTBFP(B,G,E)PDCH
DL(B,G,E)PDCH TBFs per PDCH

Steps during both active and non-active TBFs

Counts for sharing not really happening including Delayed Release
Shows higher values than in reality, compare with Data/active TBF

Includes GMM/SM
Good, but GMM/SM is a short transfer, see above

Changes/news coming
New counters in R12 counting only sharing of active TBFs

EAB-05:001190 Uen Rev A 2005-01-17 7

 Radio Link Bandwidth (RLB) counters – in reality
Frequency
B-TBF DL Radio Link bitrate Histogram
Cumulative %
12000 120.00%

10000 100.00%

8000 80.00%
# Samples

6000 60.00%

4000 40.00%
Frequency
B-TBF UL Radio Link bitrate Histogram
2000 20.00% Cumulative %
1600 100%
0 .00% 90%
1400
10

13
0

11.74
11.5

11.54

11.58

11.62

11.66

11.7

11.78

11.82

11.86

11.9

11.94

11.98

12.2

12.6

80%
1200
70%
[kbit/s] # Samples 1000 60%
800 50%
600 40%
30%
400
20%
200 10%
0 0%
0

10.8

12.4
0.8

1.6

2.4

3.2

4.8

5.6

6.4

7.2

8.8

9.6
[kbit/s]

EAB-05:001190 Uen Rev A 2005-01-17 8

Radio Link Bandwidth (RLB) counters
Σ(xx * INTxxBR(E)GPRSTBF)
Σ(INTxxBR(E)GPRSTBF)allxx Radio Link Bitrate

CS12DLACK
CS12DLSCHED * 0,02
Radio Link Bitrate

Slightly strange values on Downlink

CS 1-2 => quite good values, but sometimes >12kbit/s
CS 1-4 & EGPRS => initial ramping effects,
end effects,
averaging of RLB for long TBFs

Very unrealistic values on Uplink

Investigation ongoing, correction when found fault
See Downlink problems

EAB-05:001190 Uen Rev A 2005-01-17 9

RLB counter corrections
Σ(xx * INTxxBR(E)GPRSTBF)
Σ(INTxxBR(E)GPRSTBF)allxx Radio Link Bitrate

CS12DLACK
CS12DLSCHED * 0,02
Radio Link Bitrate

R10 AC-A7 and R11 AC-A2 (also available earlier – IP-Axx week 10)
DL/UL RLB corrections
Scheduled and Acknowledge RB counted in the same SW block.
Faulty calculated UL scheduling changed
Dropped TBFs are now counted until last acknowledged RB

Above change makes CS12DLACK not count total #RB correctly

New temporary counter SPARE3 (ot-GENGPRS) counts #ACK RBs.

EAB-05:001190 Uen Rev A 2005-01-17 10

IP Transfer Interrupt – KPI and PI level 1

(TBFDLGPRS + TBFDLEGPRS) / 6
LDISTFI + LDISRR + LDISOTH + FLUDISC = 5 – 100 minutes

FLUDISC LDISTFI LDISRR LDISOTH

(TBFULGPRS + TBFULEGPRS) / 6
IAULREL + PREJTFI + PREJOTH + PREEMPTULREL + OTHERULREL
< 1 minute

R11

IAULREL PREJTFI PREJOTH PREEMPTULREL PTHERULREL

EAB-05:001190 Uen Rev A 2005-01-17 11

IP Transfer Interrupt Downlink counters
(TBFDLGPRS + TBFDLEGPRS) / 6 Nom. = all traffic, active & non-active TBF
LDISTFI + LDISRR + LDISOTH + FLUDISC Denominator = only user data traffic

FLUDISC Almost always 0

Should step for Inter-BSC CRS shorter than 5 sec.
Increased to 15 seconds in R10 AC-A7

LDISTFI Steps OK

LDISRR Steps also for slow Inter-BSC CRS

Correction coming, see above

LDISOTH Steps OK

EAB-05:001190 Uen Rev A 2005-01-17 12

IP Transfer Interrupt Uplink counters
(TBFULGPRS + TBFULEGPRS) / 6 Nominator = all traffic, active & non-active TBF
IAULREL + PREJTFI + PREJOTH Denominator = user data traffic + some GMM

IAULREL Steps way too often

Steps also when no MS replies on Imm Ass prior
to TBF is established (~20 times too often)
Correction in R10 AC-A7 & R11 AC-A2 (or IP-Axx)

PREJTFI Steps also when no Imm Ass is sent

Correction in R10 AC-A7 & R11 AC-A2 (or IP-Axx)

PREJOTH Steps also when no Imm Ass is sent

Correction in R10 AC-A7 & R11 AC-A2 (or IP-Axx)
EAB-05:001190 Uen Rev A 2005-01-17 13
IP Transfer Interrupt Uplink counters – R11
(TBFULGPRS + TBFULEGPRS) / 6
IAULREL + PREJTFI + PREJOTH + REEMPTULREL + OTHULREL

R11 AC-A2 (or IP-Axx) required for acceptable values – not proven......

PREEMPTULREL OK (Steps for preempted TBFs with traffic)

OTHULREL Steps sometimes for similar reasons as PRETFI

Correction in R11 AC-A2 (or IP-Axx)

EAB-05:001190 Uen Rev A 2005-01-17 14

UL TBF Connection – Overview + new counter
Channel Requests: PSCHREQ

TBF Reservation phase

Accepted Channel Request: PSCHREQ-
Access Rejects
PREJTFI-PREJOTH
PREJTFI+
PREJOTH

TBF Establishment phase

TBF setup failures
PSCHREQ -PREJTFI
-PREJOTH- SPARE1 MS established on TBF: SPARE1

MS Established on TBF

Abnormally released TBFs (CV>0):

IAULREL+PREEMTULREL+ OTHULREL

Successful TBFs (not released with

CV>0):
(Underline = R11 counter SPARE1-IAULREL-
Spare1 from R10 ACA-7, PREEMPTULREL-OTHULREL
replaced in R12 by real counter)

EAB-05:001190 Uen Rev A 2005-01-17 15

Other counters and STS usage

• PCU
• GPRS/EGPRS MS Power Control
• Multislot utilization
• ...

EAB-05:001190 Uen Rev A 2005-01-17 16

 PCU counters
PCU Capacity

80
70
40
60
50
GSL Load
[%]

40
RPP Load
30
20
10
10
0
00-40 41-60 61-80 81-90 91-100 RPP PCU
Load [%] cong. cong.

100 * GSL9100
GSLSCAN Load distribution Congestion
occurences
100 * RPP9100 ALLPDCHPCUFAIL FAILMOVECELL
RPP0040 + RPP4160 + RPP6180 + RPP8190 + RPP9100 PCHALLATT ΣCELLMOVEDcell + FAILMOVECELL

EAB-05:001190 Uen Rev A 2005-01-17 18

 Live Example

GPRS/EGPRS Power Control vs GAMMA setting

Succesful UL establishment
UL Performance Indicator GAMMA dependencies IP interrupt UL
Radio Link bitrate UL
180%

160%

140%

120%
% of average

100%

80%

60%

40%

20%

0%
0 16 18 20 22 24 26 28
GAMMA

EAB-05:001190 Uen Rev A 2005-01-17 19

 100 - MUTIL12 + MUTIL22 + MUTIL13 ... MUTIL34 + MUTIL44
TRAFF2BTBFSCAN + TRAFF2GTBFSCAN + TRAFF2ETBFSCAN

MUTIL12+MUTIL13+MUTIL14
MUTIL12+MUTIL22+…+MUTIL34+MUTIL44 MS max reservable TS distribution

70.00

Multislot Utilization 60.00

50.00

Reservable TS – active (no MS class) 40.00

BSC1a

[%]
BSC9
30.00

20.00

Actual reservations made 10.00

0.00
1/1 Classed 2 TS reserveable 3 TS reserveable 4 TS reserveable

50% Reservation result, 2-4 TS capable MS

45%
90.00
40% 80.00

35% 70.00

30% 60.00

50.00 BSC1a
[%]
25%
40.00 BSC9
20%
30.00
15%
20.00
10%
10.00

5% 0.00
1 TS not max 2 TS not max 3 TS not max Max TS
0% reserved

MUTIL12+MUTIL13+MUTIL14 MUTIL22+MUTIL33+MUTIL44
MUTIL12+MUTIL22+…+MUTIL34+MUTIL44
EAB-05:001190 Uen Rev A 2005-01-17 20 MUTIL12+MUTIL22+…+MUTIL34+MUTIL44
E-PDCH Capacity
1) Sharing on B, G & E-PDCHs vs. ~TBFxLLIMIT
DLTBFPBPDCH + DLTBFPGPDCH + DLTBFPEPDCH => more PDCH capacity
DLBPDCH + DLGPDCH + DLEPDCH

2) Sharing on E-PDCHs vs. Sharing on B, G & E-PDCHs

DLTBFPEPDCH DLTBFPBPDCH + DLTBFPGPDCH + DLTBFPEPDCH
DLEPDCH DLBPDCH + DLGPDCH + DLEPDCH

> => more capacity need for E-PDCH

Difference to accept is unknown today

2) Utilization for E-TBF

MUTILEGPRS
TRAFF2ETBFSCAN
=> dedicated more E-PDCHs (or more E-PDCHs)

EAB-05:001190 Uen Rev A 2005-01-17 21

New statistics in BSS R11

Few
IP Throughput IP Transfer Interrupts IP Latency
? measurements!

Interference Capacity Mobility

RLC thp RLC thp CCCH PCU Multislot PDCH Inter RA or Intra RA & Feature Multislot ~GPRS/EGPRS All abnormally
DL UL congestion congestion utilization sharing PCU CRS PCU CRS
related utilization UL Availability released TBFs

Sharing Limited No USF HW Pre- No MS Drive

…. PDCHs on GMM MS capability
PDCHs PDCHs or TFI fault² emption PDCH capability R-PMO test
CHGR0 volume per mode

and more......
EAB-05:001190 Uen Rev A 2005-01-17 22
Work in groups (PM Detailed)
Main Tasks
• Extract the worst cells for:
– Capacity area:
– Interference area:
– Mobility area:
• What criteria have you used for each area above?
• How to separate GPRS and EDGE PDCH congestion?
• Other ideas/areas of optimization?
– Formulas, ideas and results
• What data aggregation shall be used for what analysis?

Prepare one or two slides for a presentation with above & interesting results
Please be back for presentation at ………….
EAB-05:001190 Uen Rev A 2005-01-17 23
Supportive Slides

EAB-05:001190 Uen Rev A 2005-01-17 24

 Example of GPRS/EGPRS STS in BSS

Capacity

End to End

EAB-05:001190 Uen Rev A 2005-01-17 25

 Example

STS Indications of Interference Problems:

• Sent and acknowledged Radio Blocks seen in STS:

– Total RLC throughput (UL/DL) for all users per timeslot, average per TBF,
differently calculated for GPRS CS1-2, GPRS CS1-4 and EGPRS:
CS12DLACK Σ(xx * INTxxBRGPRSTBF) Σ(xx * INTxxBREGPRSTBF)
CS12DLSCHED * 0,020 Σ(INTxxBRGPRSTBF)allxx Σ(INTxxBREGPRSTBF)allxx

# TSs or sharing does not affect this value

• Abnormally released TBFs (due to radio reasons)

LDISRR
(TBFDLGPRS + TBFDLEGPRS) * 6

EAB-05:001190 Uen Rev A 2005-01-17 26

 Example

STS Indications of Mobility Problems:

• High Cell Reselection ratio measured in the statistics:

(TBFDLGPRS + TBFDLEGPRS) / 6
– TBF minutes per inter LA/BSC CRS FLUDISC

– TBF minutes per intra LA/BSC CRS (TBFDLGPRS + TBFDLEGPRS) / 6

FLUMOVE

EAB-05:001190 Uen Rev A 2005-01-17 27

 Example

STS Indications of Capacity Problems, CCCH:

CCCH congestion risk

• Page discarded in the BTS (indicates CCCH capacity problems)
PAGPCHCONG + PAGETOOOLD > 0

• Page congestion in the BSC

TOTCONGPAG > 0

• RACH capacity is currently no limitation

– Measured RACH load: PDRAC + CNROCNT + RAACCFA
– Theoretical maximum ~40-60 RA/sec and cell

EAB-05:001190 Uen Rev A 2005-01-17 28

 Example

STS Indications of Capacity Problems, PCU:

PCU Capacity

80
70
40
60
50
GSL Load
[%]

40
RPP Load
30
20
10
10
0
00-40 41-60 61-80 81-90 91-100 RPP PCU
Load [%] cong. cong.

100 * GSLx1y0
GSLSCAN Load distribution Congestion
occurences
100 * RPP9100 ALLPDCHPCUFAIL FAILMOVECELL
RPP0040 + RPP4160 + RPP6180 + RPP8190 + RPP9100 PCHALLATT ΣCELLMOVEDcell + FAILMOVECELL

EAB-05:001190 Uen Rev A 2005-01-17 29

 Example

STS Indications of Capacity Problems, PDCH (1):

PDCH Congestion Indications
•Failure rate for PDCH Allocation Attempts
It is not possible to allocated the requested number of PDCHs.
May be a shortage, but low failure rates can be OK.
100 * PCHALLFAIL / PCHALLATT

•Average number of TBF / PDCH

This means that PDCH are shared by users (TBFs)
Compare with TBFxLLIMIT
DLTBFPBPDCH + DLTBFPGPDCH + DLTBFPEPDCH
DLBPDCH + DLGPDCH + DLEPDCH

EAB-05:001190 Uen Rev A 2005-01-17 30

 Example

STS Indications of Capacity Problems, PDCH (2):

MS Multislot Capability Measures

•Multislot Utilization
The ratio of TS received compared to TS requested
100 * (MUTILBASIC + MUTILGPRS + MUTILEGPRS)
(TRAFF2BTBFSCAN + TRAFF2GTBFSCAN + TRAFF2ETBFSCAN)

•Average maximum number of TS reservable per TBF

Indicates what MS multislot classes are used in the network
MAXTSDL / (TRAFF2BTBFSCAN + TRAFF2GTBFSCAN + TRAFF2ETBFSCAN)

EAB-05:001190 Uen Rev A 2005-01-17 31