Huawei NRT ISHO Success rate

[kyokuman]

Hi,

I have quite bad ISHO NRT success rate from 3G to 2G.

I am checking this counters and values are quite high :
Number of Timeouts of Waiting for IU RELEASE COMMAND Messages During Outgoing Inter-RAT PS Handovers Initiated by the RNC (Cell)

Do you know which timer is related to it? When does it start and expire?

I imagine it start when RNC sends CCO request to Ue.
And expire if RNC doesnt receive Iu release command from CN.

However, I couldnt find the timer itself. Which parameter is it?

Thanks for your help

[kyokuman]

Hi,

Anybody having any idea? any impression to share please?

Thanks

[agenov]

Hi,

I have quite bad ISHO NRT success rate from 3G to 2G.

I am checking this counters and values are quite high :
Number of Timeouts of Waiting for IU RELEASE COMMAND Messages During Outgoing Inter-RAT PS Handovers Initiated by the RNC (Cell)

Do you know which timer is related to it? When does it start and expire?

I imagine it start when RNC sends CCO request to Ue.
And expire if RNC doesnt receive Iu release command from CN.

However, I couldnt find the timer itself. Which parameter is it?

Thanks for your help


Under SET USTATETIMER you may see all timers related.

BR
Alex

[kyokuman]

Yes but 1 timer should start when RNC sends CCO request to Ue.

None of them describes such scenario...

[boring]

well i had a close look and i think there is no such timer in Huawei's documentation

i strongly suspect that they have a rather long hard-coded supervising timer (e.g. 30 sec or more)

[agenov]

No, you are not right. CHeck again - Number of Timeouts of Waiting for IU RELEASE COMMAND Messages During Outgoing Inter-RAT PS Handovers Initiated by the RNC (Cell).

Or in another words, after the RNC sends request for Iu release of course it needs to receive a command from CN. After particular timer expiration the RNC will release the connection by itself. So the RNC is the one initiating the release and waiting for command that is not received within particular time.

BR
Alex

[boring]

you need to be careful. the RNC does not send any iu release request for PS ISHO. signalling goes like this:

RNC --> UE: CCO from UTRAN
(UE synchs with GPRS)
CN --> RNC: Iu release command

so after sending the CCO message the RNC starts its internal timer which stops only if it receives the Iu release command

is it clear?

[kyokuman]

Yes it is correct. As I mentionned from beginning.

And if timer expires, RNC sends Iu Release REQUEST to CN.

Still wondering about which timer they mention here. Could it be RRCIuRELCMDTimer ?

[agenov]

you need to be careful. the RNC does not send any iu release request for PS ISHO. signalling goes like this:

RNC --> UE: CCO from UTRAN
(UE synchs with GPRS)
CN --> RNC: Iu release command

so after sending the CCO message the RNC starts its internal timer which stops only if it receives the Iu release command

is it clear?


This is true and like that. I just gave an example of normal release Iu release. Of course for Inter RAT is like you said and is similar procedure...

BR
Alex

[boring]

Yes it is correct. As I mentionned from beginning.

And if timer expires, RNC sends Iu Release REQUEST to CN.

Still wondering about which timer they mention here. Could it be RRCIuRELCMDTimer ?

i couldn't find this timer. any more info about this?
which section describes it?

[kyokuman]

You can find info related to that timer in :
Set calling Process state protection timers (STATETIMER)

Got not more info.
However : RNC should send Iu Release request to CN after expiration of TRELOCOverall (3GPP).
But I can't find corresponding Huawei timer...

[boring]

Hi guys
After spending some time I have found the following:

expected signalling

UE <-- RNC (Cell change order from UTRAN)
UE synchs with GPRS
RNC <-- SGSN (SRNS context request)
RNC --> SGSN (SRNS context response)
RNC <-- SGSN (SRNS data forward command)
RNC <-- SGSN (Iu release command)
RNC --> SGSN (Iu release complete)

When the RNC sends the CCO message, it starts the T309 timer (= 5sec)
If the timer expires, the UE is considered as either lost or successfully moved to GPRS. The RNC is free to release radio resources but maintains the UE context until it will receive the SRNS context request message

The question of course is for how long the RNC would wait for the SRNS context request message

Unfortunately Huawei documentation is not clear on this at all. I strongly suspect that the supervising timer for this procedure could be one of (unless it is hard-coded of course):

RELOCCMDTMR (=10 sec)
RELOCIURELCMDTMR (=20 sec)
RELOCDATAFWDTMR (=20 sec)

[kyokuman]

Hi,

Yes not clear at all. I have asked Huawei guys who asked their Rnd.

Apparently it is RelocIuRelCmdTmr, which is set here to 45s.

Start : CCO sent to Ue
Stop : When RNC receives Iu Release command

As we have low ISHO NRT sucess rate, Huawei proposes 60s ! I disagree...
And the thing is that I cant find the corresponding NSN timer...
It could be TrelocOverall, but this concerns CS only. I'm not sure...

[agenov]

But you already have the answer :). For how long it should wait for command release message (RELOCIURELCMDTMR=20s). But usually default value is 45s.

BR
Alex


Hi guys
After spending some time I have found the following:

expected signalling

UE <-- RNC (Cell change order from UTRAN)
UE synchs with GPRS
RNC <-- SGSN (SRNS context request)
RNC --> SGSN (SRNS context response)
RNC <-- SGSN (SRNS data forward command)
RNC <-- SGSN (Iu release command)
RNC --> SGSN (Iu release complete)

When the RNC sends the CCO message, it starts the T309 timer (= 5sec)
If the timer expires, the UE is considered as either lost or successfully moved to GPRS. The RNC is free to release radio resources but maintains the UE context until it will receive the SRNS context request message

The question of course is for how long the RNC would wait for the SRNS context request message

Unfortunately Huawei documentation is not clear on this at all. I strongly suspect that the supervising timer for this procedure could be one of (unless it is hard-coded of course):

RELOCCMDTMR (=10 sec)
RELOCIURELCMDTMR (=20 sec)
RELOCDATAFWDTMR (=20 sec)

[boring]

OK it is good to know that they use a completely different timer for this procedure :lol

anyway. now i completetly agree with you. don't increase the timer unless there is something very wrong at the GPRS side.

the NRT ISHO failures are primarily due to the counter we are talking about or due to CCOFU Failure?

[kyokuman]

Unfortunately due to this counter.

[boring]

some additional information:

RSCP and EcNo thresholds for event 2d measurements?
what is the value of T309?
what is the typical size of the inter-system neighbouring lists?
BSIC verification is on for GPRS?

the problem is widespread across the network or is localised?

[kyokuman]

RSCP and EcNo thresholds for event 2d measurements?

RSCP : -110dBm (note that it was -112dBm before swap but adjusted to have same amount of HO)
Ecno : Not activated for PS

what is the value of T309? 5s

what is the typical size of the inter-system neighbouring lists? Average 15

BSIC verification is on for GPRS? Yes it is.

the problem is widespread across the network or is localised?
After Swap NRT ISHO success rate = 97%
Before swap : 99%
widespread. Not huge degradation but still

[boring]

ok judging from the current parameters, and the fact that you were using NSN, i believe that the only main difference is the fact that BSIC verification is on for NRT.

NSN does not use BSIC verification for NRT unless there are two cells with the same BCCH in the combined monitoring list

for me BSIC verification delays unnecessarily NRT cells changes and perhaps this is the reason for the difference. but it is very difficult to prove because i think Huawei has a global BSIC verification switch for both RT and NRT calls

[kyokuman]

Hi,

Thanks for help!

I also just compared other parameters such as Max Compressed mode duration timer, or period for event reporting, timetotrigger for HO decision (not for 2d event).
And the values are very different from NSN to Huawei.
MR reporting is every 1 second for Huawei, Half a sec for NSN
Max CM measurement period is 1 minute for Huawei, a few seconds for NSN...

I'm planning to proceed with some optim of those timers and parameters.

If you have any good info for me, thanks to let me know

[dk2000]

RSCP and EcNo thresholds for event 2d measurements?

RSCP : -110dBm (note that it was -112dBm before swap but adjusted to have same amount of HO)
Ecno : Not activated for PS

what is the value of T309? 5s

what is the typical size of the inter-system neighbouring lists? Average 15

BSIC verification is on for GPRS? Yes it is.

the problem is widespread across the network or is localised?
After Swap NRT ISHO success rate = 97%
Before swap : 99%
widespread. Not huge degradation but still


Which is the vendor of 2G?

[kyokuman]

Both Huawei and Siemens