[Openbts-discuss] Fwd: GPRS protocol issue logs (MS side)

[Michael I. <mic...@ra...>]

Just noticed these did not go out to the list. Forwarding the first and will reply with other entries in the thread shortly.


> Begin forwarded message:
> 
> Date: November 29, 2014 at 4:17:50 PM GMT+1
> From: Sipos Csaba <sip...@kv...>
> To: <mic...@ra...>, Tom Tsou <to...@ts...>
> Cc: <ope...@li...>
> Subject: GPRS protocol issue logs (MS side)
> 
> Hey guys,
> 
> I just wanted to give you some new findings about the always returning issue with GPRS.
> 
> Today I had some time to analyze the issue, here are the results:
> 
> In almost every case, the GPRS instability or sometimes even the complete loss of service is happening like this:
> 
> EVENT  EVENT_GPRS_MAC_MSG_SENT        14:39:47.629   Message: Packet Downlink Ack/Nack  Channel: 0
> EVENT  EVENT_GPRS_MAC_MSG_RECEIVED    14:39:47.725   Message: Packet Uplink Ack/Nack  Channel: 0
> EVENT  EVENT_GPRS_MAC_MSG_SENT        14:39:47.725   Message: Packet Control Acknowledgement  Channel: 0
> EVENT  EVENT_GPRS_TBF_RELEASE         14:39:48.156   TBF Release Type: DL_TBF_NORMAL
> EVENT  EVENT_GPRS_MAC_MSG_SENT        14:39:50.045   Message: Packet Measurement Report  Channel: 0
> 
> !!! ERROR EVENT !!!
> 
> EVENT  EVENT_GPRS_TIMER_EXPIRY        14:39:53.003   GPRS Timer ID: T3180
> EVENT  EVENT_GPRS_TBF_RELEASE         14:39:53.003   TBF Release Type: UL_TBF_ABNORMAL
> EVENT  EVENT_GSM_L1_STATE             14:39:53.015   L1_IDLE_MODE
> EVENT  EVENT_GSM_L1_STATE             14:39:53.024   L1_RANDOM_ACCESS_MODE
> EVENT  EVENT_RANDOM_ACCESS_REQUEST    14:39:53.032   1 31601 GMAC_ONE_PHASE_ACCESS
> EVENT  EVENT_GSM_MESSAGE_RECEIVED     14:39:53.212   IMMEDIATE_ASSIGNMENT
> 
> As you can see from time to time at least one RLC/MAC block does not get Acked/Nacked before the timer (T3180) expires. 
> Most of the times, this happens right after a Packet Meas. Report, but not always.
> When this timer expires, the MS sets UL_TBF_ABNORMAL and tries with a Random Access (as the specification requires).
> 
> To be more specific, the T3180 timer should stop when detecting an assigned USF value on assigned PDCH, so it seems the Ack/Nack cannot happen because OpenBTS does not send an USF allocation for the MS.
> 
> The test was a constant ping from the MS to the SGSNTUN interface.
> 
> As you will see in the full log (attached), after 30 seconds the MS was able to continoue the GPRS session, but in 50% of the cases this is not happening, I will try and give you a log about that, too.
> 
> The MS is 1 meter away from the BTS (Ettus B200), with -50 RXQUAL and -35 to -40 Uplink RSSI. FER is 0%, no interference on this channel.
> 
> I will continue with the protocol traces and share my findings with you.
> 
> I hope I can help at least with locating these issues.
> 
> Regards,
> Csaba