You are on page 1of 10

TeliaSonera Sweden optimisation update

ongoing project(no final report available)

For internal use only


1
Nokia Siemens Networks

Introduction
The TeliaSonera Sweden optimisation task force (TSSOTF) currently
focuses on

1. Optimisation of the eNB parameter values, with respect of KPIs from the
system performance evaluation agreement towards the customer, and
those reported to upper managerial layers.
2. Follow-up of outstanding faults that can impact the aforementioned KPIs.
This includes
1. Specific UE issues
2. Specific core network issues
3. Internal eNB faults that can be determined from alarms/PM data/drive
testing analysis.

For internal use only


2
Nokia Siemens Networks

What has been done, so far


Summary of software update and parameter changes

1.
2.
3.
4.
5.
6.
7.
8.
9.

Upgrade to RL20: valid from March 8th 2011 in the nine golden cluster sites.
First parameter alignment with default golden parameters and/or some parameters
previously tuned for Gothenburg: valid from March 16th 2011.
Minor parameter change in the golden cluster, tweak for TP during drive testing associated
to system performance evaluation (SPE), maxNrSymPdcch 32 (fine insofar that the
network has very light load)
Link parameters changed in the golden cluster, valid from 25th/March (nCqiDtx,t310,)
PRACH parameters + HO correction valid from 30th/March in the golden cluster
Network-wide alignment with golden recommendations, valid from April 6th.
Network-wide PRACH parameters tested in the golden cluster, valid from April 6th.
Link+PRACH parameters changed in a test cluster, valid from April 7th (nCqiDtx,t310,)
Additional testing for feedback to GMC: valid from April 20th, Lindholmen: taCmdMaxRetry
10->3 and Brunnsparken: taOffScheMarg 40->20
Date
Golden Cluster
8-Mar-11 Upgrade to RL20 PCD 1.1
9-Mar-11
16-Mar-11

For internal use only


3
Nokia Siemens Networks

Gothenborg

Whole Network

Upgrade to RL20 PCD 1.1


Upgrade to RL20 PCD 1.1

SW
updates

Summary of activities for c.w. 15, 16, 17


1. Discussion about default values for vendor file parameters, and how to deploy
the parameter values currently modified in the golden cluster, to the rest of the
network.
2. Discussion about the root cause of the radio link drops, which cause bad KPIs
with the default parameters.
3. Discussion about alignment of RL20 default values with recommended values
for RL10, agreed during the maintenance phase.
4. Troubleshooting for the outstanding problems affecting the KPIs.
1. RRC setup failures due to missing messages from the UE.
2. RRC drops due to missing messages from EPC.
3. FSP crashes.
5. Monitoring of the KPIs.
6. Troubleshooting of miscellaneous issues:
1. Occasional problems in the NetAct result in missing data in the Oracle
database. The data is still available in the iOMS.
2. Three sites crashed after vendor-configurable parameter updates.
For internal use only
4
Nokia Siemens Networks

Deployment of modified parameter values to the


rest of the network
1. There is no automated way to modify the vendor files while keeping the current
2.
3.
4.
5.

software version. Manual change of the vendor files is required, which is errorprone.
A custom build for PCD 1.1 was produced, so that the parameters could be
deployed network-wide upon the software upgrade.
However, the customer chose not to upgrade to PCD 1.1 before Easter.
After Easter, the upcoming release of PCD 2.2 makes the customer reconsider
about upgrading to PCD 2.1. Upgrade to PCD 2.2 may be done instead.
Most of the modified vendor file parameter changes have been agreed upon, to
be deployed in PCD 2.2. Exception made for taCmdMaxRetry and
taOffScheMarg. These are currently being tested, each one in an individual
site.

For internal use only


5
Nokia Siemens Networks

Discussion about the root cause of the radio link


drops, which cause bad KPIs with the default
parameters
1. Traces have been delivered to development, to analyze the current situation of
2.
3.
4.

RL20: both eRAB drop ratio and RRC success ratio have bad KPI values with
the default parameter values.
A discussion on the analysis of the traces followed.
The increase in radio link drops can be explained partially by a different
behaviour of the eNB, upon arrival of RRC connection re-establishment
requests. This does not account for all the drops, though.
Development has requested more extensive tracing. A large TTI trace
collection campaign is being organized.

For internal use only


6
Nokia Siemens Networks

Monitoring of the KPIs: eRAB drop ratio in the


golden cluster
Drops in
erab_DR 100427
18 04 2011
22.78
4
19 04 2011
8.97
6
20 04 2011
24.36
13
21 04 2011
11.06
28
22 04 2011
17.74
2
23 04 2011
27.94
0
24 04 2011
21.33
1

100428
0
0
1
1
0
0
0

100435
0
0
1
0
0
0
0

100438
0
0
0
0
0
0
0

100445
2
0
45
17
25
29
28

Sporadic
drops in
Gtaplatsen
enb_rel_rnl Succ. eRAB setup Total eRAB releases
18 04 2011
36
155
158
19 04 2011
27
298
301
20 04 2011
133
544
546
21 04 2011
52
464
470
22 04 2011
33
186
186
23 04 2011
38
137
136
24 04 2011
32
149
150

For internal use only


7
Nokia Siemens Networks

100448
0
0
5
0
0
0
0

100480
0
0
6
0
0
0
2

100493
0
0
0
0
0
0
0

100494
30
21
62
6
6
9
1

Kaserntorget will
be investigated
from the TTI +
Emil traces

Additional
drops in
Svenska
Mssan, to
be checked

Traffic has been somewhat


lower last week (about 18%
less eRAB setups compared to
the previous week)

Parameter changes: test of taCmdMaxRetry 103


and taOffScheMarg 4020
Vendor file
parameters
changed. Large
decrease in nr of
drops

Lindholmen: large
variability in the
drops, no evident
change.
Brunnsparken:
situation may have
improved, but
there are too few
drops to be
certain,.

Lindholmen (100468):
taCmdMaxRetry 103
Brunnsparken:
taOffScheMarg 4020

60
50
40
30
100443
20

100468

10
0
28 04 2011

27 04 2011

26 04 2011

25 04 2011

24 04 2011

23 04 2011

22 04 2011

21 04 2011

20 04 2011

19 04 2011

18 04 2011

17 04 2011

16 04 2011

15 04 2011

14 04 2011

13 04 2011

12 04 2011

11 04 2011

10 04 2011

09 04 2011

08 04 2011

07 04 2011

For internal use only


8
Nokia Siemens Networks

PM data study: UL SINR distribution RL 10 / RL 20

No significant change found.


UL CL PC disabled,
ulpcIniPrePwr = -90 dBm,
ulpcRarespTpc = 2 dB in
both releases.
It has been discussed that a
lower value could be used,
along with a higher value for
ulpcRarespTpc. This could
be tested in the field
verification cluster in Espoo.

For internal use only


9
Nokia Siemens Networks

PM data study: UL MCS distribution


MCS usage for April 19th, cells with PDCP UL vol > 30 MB
70

60

Rough agreement between


the MCS distribution / TP
distribution (maximum TP per
cell per 15 minutes), and the
drive testing TP curve.

Usage [%]

50

40

30

20

10

10

UL TP v/s
RSRP per
MCS

For internal use only


10
Nokia Siemens Networks

15
MCS

20

25

30

You might also like