PDA

View Full Version : RRC Admission Control Failures - Troubleshooting Techniques



maq17
2012-02-21, 10:52 AM
Dear experts

I have a problem dealing with admission control. The 3G network is NSN and i use also Traffica for real time monitoring in some KPI's related to RRC setup and access

First, I have one sector with about 6000 RRC attempts (400 for RT and 4000 for NRT, rest unspecified) per hour. In that hour, the failure event is downlink overload (as i can see in Traffica). I tried to correct this failures reducing the CPICH power from 33dBm to 30dBm, but after that, same problem in RRC occur but this time due to lack of codes.

No problems related to Channel Elements or Congestion in Iub were detected. Also I tried to reduce the PDSCH code set limiting from 15 codes to 5 codes and the result is the same. Curious concern is that I have too much HSDPA users (almost 50 in the sector but with no user license blocking) and HSDPA traffic about only 1.8Mbps in one hour

Do you have any idea for this problem?

jan74
2012-02-21, 07:03 PM
Dear experts

I have a problem dealing with admission control. The 3G network is NSN and i use also Traffica for real time monitoring in some KPI's related to RRC setup and access

First, I have one sector with about 6000 RRC attempts (400 for RT and 4000 for NRT, rest unspecified) per hour. In that hour, the failure event is downlink overload (as i can see in Traffica). I tried to correct this failures reducing the CPICH power from 33dBm to 30dBm, but after that, same problem in RRC occur but this time due to lack of codes.

No problems related to Channel Elements or Congestion in Iub were detected. Also I tried to reduce the PDSCH code set limiting from 15 codes to 5 codes and the result is the same. Curious concern is that I have too much HSDPA users (almost 50 in the sector but with no user license blocking) and HSDPA traffic about only 1.8Mbps in one hour

Do you have any idea for this problem?

Admission Conrtrol is difficult to understand in NSN network. In theory, you should get RRC Connection Setup Fail AC when the cell is in Cell Overload. However, it can be more complex than that and the stats are not the clearest. Check counters PTXTOT_DENOM_4 (M1000C23) and prxtot_denom_4 (M1000C9) for the number of samples during the busy hour when the cell is overloaded. There are 7,200 samples every hour so if you see many samples in this class, you have serious overload. In theory when you have overload, you will get RRC_Conn_Setup Fail_AC (m1001C3).

At this stage, you need to check your thresholds. If prxtot_denom_4 (M1000C9) is worse, then you need to check PrxTarget, PrxTargetPSMax, PrxTargetPSMin and PrxMaxTargetBTS. If you have HSUPA active, these thresholds will be used for Ps R99. If those are set to the insane NSN default values(4dB, 6dB etc), your cell will congest immediately.

If PTXTOT_DENOM_4 (M1000C23) is the culprit, then check the available DL Tx Power. Reducing your CPICH should have helped and I cant see how you got Code congestion after that. The code congestion must be due to having reducedthe PDSCH Code set. I think you should go back to 15 on that anjd then go back to 30dBm CPICH to see what happens.

Finally, youre getting DCH_SEL_MAX_USERS_HSDPA? Have you got enough schedulers installed? Is there 1 scheduler per BTS or is there one per sector? You need to look at that. Also, have you got HSPA 72 Users enabled?

maq17
2012-02-22, 12:57 AM
Admission Conrtrol is difficult to understand in NSN network. In theory, you should get RRC Connection Setup Fail AC when the cell is in Cell Overload. However, it can be more complex than that and the stats are not the clearest. Check counters PTXTOT_DENOM_4 (M1000C23) and prxtot_denom_4 (M1000C9) for the number of samples during the busy hour when the cell is overloaded. There are 7,200 samples every hour so if you see many samples in this class, you have serious overload. In theory when you have overload, you will get RRC_Conn_Setup Fail_AC (m1001C3).

At this stage, you need to check your thresholds. If prxtot_denom_4 (M1000C9) is worse, then you need to check PrxTarget, PrxTargetPSMax, PrxTargetPSMin and PrxMaxTargetBTS. If you have HSUPA active, these thresholds will be used for Ps R99. If those are set to the insane NSN default values(4dB, 6dB etc), your cell will congest immediately.

If PTXTOT_DENOM_4 (M1000C23) is the culprit, then check the available DL Tx Power. Reducing your CPICH should have helped and I cant see how you got Code congestion after that. The code congestion must be due to having reducedthe PDSCH Code set. I think you should go back to 15 on that anjd then go back to 30dBm CPICH to see what happens.

Finally, youre getting DCH_SEL_MAX_USERS_HSDPA? Have you got enough schedulers installed? Is there 1 scheduler per BTS or is there one per sector? You need to look at that. Also, have you got HSPA 72 Users enabled?

Thanks for your quick answer

The problem is in downlink, so all the counters related to uplink are not the problem (PRXTOT_DENOM_4 or AVERAGE_UPLINK_LOAD). I change the parameters PRxTarget and PRxTargetPSxxx as the uplink interference and rejections for HSDPA due to return channel increases and works OK, thanks.

I also dont have problems with DCH_SEL_MAX_USERS_HSDPA. In that case I would have troubles with code congestion, so the licenses are well distributed in all sectors for HSDPA with 1 scheduler per sector and HSUPA with 72 users

As I could read, PDSCH Code Set reduces the amount of codes for HSDPA per connection (If I am wrong, please correct me). After reducing power, code congestion appeared in Traffica, maybe because I have many enterprise buildings in front of the sector, so I change the code set with 2,4,5 codes available per connection and the code congestion counter was reduced only for the first loaded hour (we have about 3 loaded hours). For the next two hours, code congestion appeared

I'm going to try with the counters for downlink power. Based on your experience, for a 20W amplifier, how percentage of power I should reserve for Rel99 or how can I control it?

Regards

jan74
2012-02-23, 01:24 AM
I dont understand what you mean by "Curious concern is that I have too much HSDPA users (almost 50 in the sector but with no user license blocking) ". thats why I assumed you had DCH_SEL_MAX_USERS_HSDPA. By the way, if you do have DCH_SEL_MAX_USERS_HSDPA, you do not necessarily have Code Congestion. If you have Max Users, with already 72 users feature active, then you need to consider adding another carrier and another scheduler.

For the code congestion, we have never had a problem with it even on very busy cells. We have the feature HSDPA 15 Codes enabled with 15 HS-PDSCH codes. I cant see how reducing the codeset can help as its supposed to be dynamic anyway....Have you set the codeset to 5? As far as I know, the HSDPA Dynamic code feature is disabled by setting to only 5 codes.? Have you enabled 4 users per TTI? That means having 4 HS-SCCH channels available. This would help you.

For DL power, our setup is slightly different as we try to keep HS on one carrier and R99 on another. In areas where we have that setup, we only allow 40dBm for HSDPA. We also prioritise R99.

patek1968
2012-06-23, 06:12 AM
Why not try cell reselection?