Re: [Openbts-discuss] BCCH power break-in

[Harald K. <kr...@ii...>]

On 12/21/2010 08:09 PM, Sylvain Munaut wrote:
> Hi,
>
>> Unfortunately I notice a break-in of the TX power in irregular intervals
>> as shown in the figure in [1]. The length of the break-in's
>> is around 2.154 ms or multiples of this value.
>
>> Has someone also noticed such a behavior, or someone knows where it
>> comes from, how to prevent it?
>
>
> Yes I've noticed the same.
>
> Do you get "dumping STALE bursts" message in the TRX logs ?  (not
> those are the very beginning in the first 10 secs or so. Those are
> normal).
>
> I tried the real time priority but that didn't help much :(
>
>
> Cheers,
>
>      Sylvain

Hi,

changing the process priority reduces the probability of occurrence of a 
break-in slightly. In fact on our 2.4 GHz P4 machine OpenBTS uses 30 
percent of the CPU power. By applying heavy CPU loads using the stress 
command, the signal breaks down immediately and ramps up after releasing 
the stressing.

During the break-in lines like the following are written in the TRX log:

1293089329.9532 NOTICE 3077458800 Transceiver.cpp:146:pushRadioVector: 
dumping STALE burst in TRX->USRP interface

It seems like mTransmitPriorityQueue is not updated in time.


Any suggestions to handle this?


Cheers,

Harald