PS IRAT & CS DCR Improvement - U900

[khurrambilal01]

Hi All,

We are having a U900 + U2100 NW with Huawei RAN 14.0, PS IRAT SR is very low for U900 sites, normally its like 90-91%.

1) How to improve PS IRAT, because we have already optimized HSPA/R99 2D/2F and triggering parameters.
2) How to improve CS DCR on U900 layer because it's coverage is more than U2100 and even GSM 900 because of UL noise difference b/w GSM & WCDMA? currently CS DCR is around 0.9%?

We have tried some parameters like MaxRLPower for CS AMR and Maximum DCH rate for multiRAB AMR call, any other parameters? Suggestions?

[khurrambilal01]

Hi Experts,

Any suggestions? Reputation will be added for helpful support.

[amadeo]

hi bilal,
I am not Huawei engineer but i can help you as common sense.
Improve PS IRAT:
1. Make the triggering event harder. (lower the threshold of RSCP or EcIo so that it will maintain to keep the call in the band). This will prevent pingpong or unnecessary IRAT.
2. Make the target cell qualification stricker (increase the minimum RSCP as target into -95 dBm or anything else you have in huawei).

Improve CS DCR:
1. Make sure no overshooting and have a balance uplink & downlink. Check if your UL_Out_of_Synch couonter is high but your SHO failure is low. This indicate you have limitation on uplink. This can only be solved by reducing your coverage either by downtilt or reduce CPICH.
2. CHeck your PRACH plot, make sure the distance until 98% of your traffic area has all neighbors defined.
3. Check your SHO OH. If they are low (< 20%), that means you have coverage gap. Try to uptilt your neighboring cell and also check the serving cell. If they are high (> 40%) and your prach plot is far (>10 km), downtilt your server.
4. Increase your T313

Hope this will help.

cheers,

amadeo

[khurrambilal01]

Hi amadeo (http://www.finetopix.com/member.php?9261-amadeo);
Thanks for the reply.

I have already changed CS 2D/2F to -14/-12 with hysteresis zero, CS RSCP 2D/2F -97/-95; PS 2D/2F is -16/-14 and RSCP is -115/-110.

For interfreq Target EcNo threshold is set to -10 and RSCP is -95 while for GSM target RSSI is -90...

NinsynInd is changed to 3 from 5 and outsyncind is default value 50.

Still CS DCR is 0.8 and CS IRAT is 97 while target is 97.5, PS IRAT is 91% while target is 95%.

[Syed Fahad]

Khurram Sb,

Do we know the IRAT Failures are in Preparation or Execution Phase.
DL / UL TBF Congestion on 2G can impact the Preparation Failures .
Some times very low Values for PDCH Rate at 2G End can give rise to TBF Congestion impacting PS IRAT Prep Failures.
While External Inference at 2G End also impacts IRAT Execution Failures.
In NSN we have tried Multiple BSIC Identification Feature in which BSIC Identification of more than One GSM Neighbor is performed that lead to slight improvement in CS IRAT Success Rate.
For CS DCR, If T314 > 8s / 10s than 0.8 is way too much provided no External Interference issues.

[dtvt2]

Hi
- Enable BSIC verify
- Planning frequency 2G

[khurrambilal01]

Hi
- Enable BSIC verify
- Planning frequency 2G


BSIC verfiy switch is already enabled and we have recently done AFP for 2G...

[khurrambilal01]

Khurram Sb,

Do we know the IRAT Failures are in Preparation or Execution Phase.
DL / UL TBF Congestion on 2G can impact the Preparation Failures .
Some times very low Values for PDCH Rate at 2G End can give rise to TBF Congestion impacting PS IRAT Prep Failures.
While External Inference at 2G End also impacts IRAT Execution Failures.
In NSN we have tried Multiple BSIC Identification Feature in which BSIC Identification of more than One GSM Neighbor is performed that lead to slight improvement in CS IRAT Success Rate.
For CS DCR, If T314 > 8s / 10s than 0.8 is way too much provided no External Interference issues.

Hi Syed Fahad (http://www.finetopix.com/member.php?68824-Syed-Fahad) ;

IRAT failures are in Execution phase with reason physical channel failure.
BSIC verfiry switch is already enabled, External interference on 2G is only present for two airport sites, rest all are clear, Although one thing that i think can be the cause is NBR list truncation because we have defined many NBRs for IRAT as well as IF (U2100 F1/F2 And U900 F3).

T313 value is 3sec, but this timer is RNC level and we have good performance for U2100 with same timers in same RNC but U900 has higher DCR.

[Syed Fahad]

Being X Huawei, i understand that huawei Management wont take it as an excuse, but as matter of fact after deployment of third carrier on 900Mhz, your UMTS 900Mhz Traffic and coverage might be increased and improved. The Routine coverage related VIP Customer complaints on Naster would be reduced. But i am of the view that comparing U2100 KPIs with U900Mhz KPI is not justified.. U900Mhz being on lower frequency will be having high penetration power and with respect to U2100 and vice vera. Noise floor on U900Mhz will be high as result of that with compare to U2100, as it will be taking higher share of traffic.
Any ways coming to PS IRAT Issue, can we make sure that is there any feature in Huawei with Force Decode BSIC Option for NRT. As per my understanding BSIC is not decoded for NRT IRAT HO/ Cell Changes, its only for CS IRAT.

[khurrambilal01]

Being X Huawei, i understand that huawei Management wont take it as an excuse, but as matter of fact after deployment of third carrier on 900Mhz, your UMTS 900Mhz Traffic and coverage might be increased and improved. The Routine coverage related VIP Customer complaints on Naster would be reduced. But i am of the view that comparing U2100 KPIs with U900Mhz KPI is not justified.. U900Mhz being on lower frequency will be having high penetration power and with respect to U2100 and vice vera. Noise floor on U900Mhz will be high as result of that with compare to U2100, as it will be taking higher share of traffic.
Any ways coming to PS IRAT Issue, can we make sure that is there any feature in Huawei with Force Decode BSIC Option for NRT. As per my understanding BSIC is not decoded for NRT IRAT HO/ Cell Changes, its only for CS IRAT.

Yes U900 has improved coverage, Traffic and penetration. RTWP is relatively high as compared to U2100 layer, Alot of CS traffic has been captured by U900 + HSDPA traffic, Some cells are bursting. 20-30% revenue has increased.

I can't find any feature which can force BSIC verification but there is a parameter which can delay/disable handover when BSIC is not verified.I would try this parameter but risk is increase in call drops...

[omarGSMUMTS]

coul you share some doc . thank you!

[rumy]

Check the No Cell Found Ratio; it gives you an idea about :
-Wrong GSM Config
-No GSM coverage

Also, you may have restricted International Roamers who do not have access at GSM (this is my case, GSM is used from another operator :) )

Hope it helped!

[khurrambilal01]

Check the No Cell Found Ratio; it gives you an idea about :
-Wrong GSM Config
-No GSM coverage

Also, you may have restricted International Roamers who do not have access at GSM (this is my case, GSM is used from another operator :) )


Hope it helped!

Around 30% of attempts end with no GSM cell found for IRAT handover which I can optimize by reducing the target RSSI required threshold for GSM.

Secondly do you mean to say that international roamers which don't have access to GSM because of coverage are all carried over U900?? U900 with same power as G900 has normaly 50% more coverage area.

[litium1000]

If all U900 sites have collocated GSM900, PS IRAT HO makes no sense. U900 provides better coverage and PS user experience than G900. By allowing PS IRAT HO you will just move the drops in GSM.
Pay attention at the number of IRAT neighbors. In Compress Mode, the UE can only measure 8 BCCH's in a gap. It needs 3 measurements per each BCCH for filtering and needs another gap to confirm BSIC. Even if you have less than 8 GSM neighbors it takes 3-5 seconds for a CS IRAT HO. If your IRAT neighbor list is full, for sure it will need more time to measure all BCCH's. If the radio environment is degrading rapidly, you will have a drop before making the IRAT HO.

To have less out of sync radio links , use Ninsync=2, and NoutofSync=200

Also, put a trace in UU interface of a cell, and verify measurement control messages for 2d/2f thresholds. Make sure that the values that you set in SET UINTERRATHOCOV are the ones sent to the mobile. Check for each service: CS only, PS R99, HSDPA and multiRAB. You might have some suprises...

[khurrambilal01]

If all U900 sites have collocated GSM900, PS IRAT HO makes no sense. U900 provides better coverage and PS user experience than G900. By allowing PS IRAT HO you will just move the drops in GSM.
Pay attention at the number of IRAT neighbors. In Compress Mode, the UE can only measure 8 BCCH's in a gap. It needs 3 measurements per each BCCH for filtering and needs another gap to confirm BSIC. Even if you have less than 8 GSM neighbors it takes 3-5 seconds for a CS IRAT HO. If your IRAT neighbor list is full, for sure it will need more time to measure all BCCH's. If the radio environment is degrading rapidly, you will have a drop before making the IRAT HO.


Hi litium1000 (http://www.finetopix.com/member.php?14508-litium1000),

Thanks for the reply, All U900/G900 sites are co-located and i understand U900 coverage is better than G900 but EcNo can get bad not only on U900 borders even near to site because of Pilot pollution in some areas so we need to keep IRAT neighbours and most of IRAT attempts are EcNo based...

Secondly you are right about the time it takes to filter IRAT neighbors for CS but there is no other way possible for CS DCR improvement because once i changed 2D/2F EcNo from -16/-14 to -14/-12 i observed 20% call improvement.

Now I have changed it to -12/-10 and CS DCR is further improved although IRAT per call has increased from 20% to 30%. I have disabled IF HHO for CS so that no users in CS mode can go to U2100 from U900 because it will eventually end with IRAT call and reduce my MOS/increase interference due to compressed mode.

[litium1000]

I have just edited my last post with some sugestions..
It might be a good idea also not to allow inter-freq CS from U2100 to U900, and send them directly to IRAT. This could help in your scenario if you have many CS interfreq U2100 to U900, because this type of calls only be counted as drops in U900..they are established in U2100 cells, and possibly dropped in U900 :)). In calculating DCR in U900 you count drops/established in U900 band

[khurrambilal01]

I have just edited my last post with some sugestions..
It might be a good idea also not to allow inter-freq CS from U2100 to U900, and send them directly to IRAT. This could help in your scenario if you have many CS interfreq U2100 to U900, because this type of calls only be counted as drops in U900..they are established in U2100 cells, and possibly dropped in U900 :)). In calculating DCR in U900 you count drops/established in U900 band

Normally the dropped call during handover phase are counted on source cell that why we always normally monitor outgoing handovers.

Changing In/Out sync Indication timers seems not right to me because instead of resolving the problem we are trying to hide it by changing timers that too very high values.

[litium1000]

Normally the dropped call during handover phase are counted on source cell that why we always normally monitor outgoing handovers.

Changing In/Out sync Indication timers seems not right to me because instead of resolving the problem we are trying to hide it by changing timers that too very high values.

- I was talking about overall CS drop call rate in U900 band
-Changing InSync/Out Sync makes no difference for PS. For sure, even with this timers, user experience will be far better than GSM. You might have some problems with CS..but not necessary.

In my opinion, your strategy is your problem. It really makes no sense making HO from U900 to G900 if you have continuous layers. If you have no choice in changing the strategy, try to be more aggressive at cell level. Find worst cells, use agressive cell level EcNo/RSCP interRAT parameters, while trying to improve EcNo in the U900 cells.

[khurrambilal01]

Hi litium1000 (http://www.finetopix.com/member.php?14508-litium1000),
If we keep users to U900 layer and don't push them to G900/1800 even when EcNo/RSCP is bad, KPIs for U900 are going to be too bad and we have targets :)
Secondly U900 load is already high so keeping CS users when the coverage (EcNo) isn't good will drop KPIs as well as increase traffic.

We have to balance G900/1800, U2100 and U900 KPIs and traffic as well as provide coverage to new users which are not served by U2100...

Changing Insyn/Outsync timers will affect CS only that's true but changing timers will reduce call drops in statistics not on-ground.... (Users will still face dropped calls and instead of waiting for too long they will finish the call themselves)

[litium1000]

You can try to offload U900 to U2100 through idle reselection (if you haven't done it by now). Prioritize reselection to 2100 Band from U900 and G900. Keep idle and connected UE's as much as possible in U2100. Do not allow CS interfreq from U2100 to U900, send the calls directly to IRAT. If U2100 becomes overloaded, expand 2nd carrier (probably you have more spectrum available in 2100 than 900 band).

Good luck!

[khurrambilal01]

You can try to offload U900 to U2100 through idle reselection (if you haven't done it by now). Prioritize reselection to 2100 Band from U900 and G900. Keep idle and connected UE's as much as possible in U2100. Do not allow CS interfreq from U2100 to U900, send the calls directly to IRAT. If U2100 becomes overloaded, expand 2nd carrier (probably you have more spectrum available in 2100 than 900 band).

Good luck!
Hi litium1000 (http://www.finetopix.com/member.php?14508-litium1000),

Idle mode UEs are already on U2100 (IdleSintersearch = 9 which means EcNo <0) then UE will search for interfreq cell, so it always keeps searching for U2100.
According to DT 30% UEs are on U900 in idle mode where the coverage for U2100 is weak.

CS based interfreq HO is disabled for all U2100/U900, We have F1/F2 in U2100 and F3 on U900.

PS 2D/2F is -115/-110, -16/-14 which i guess is pretty aggressive to keep user on U900/U2100.

[mohyedeen_alkousy]

@khurrambilal01 (http://www.finetopix.com/member.php?65308-khurrambilal01)

PS IRAT SR:
I think the issue is in the TBF establishment in the target GSM Cell , if u improve it , the PS IRAT will improve also. changing any parameter in U900 will not resolve the root issue, it will either hide the problem or create another problem. beside that its better to keep the user on U900 as long as there no overload on it.




CS DCR:
best solution is to figure out why EcN0 in U900 is bad and try to improve it , check overshooting cells,hardware issues ,neighbors , wrong configuration.