[Openbts-discuss] Jumps in the RX packet timestamps ?

[Sylvain M. <24...@gm...>]

Hi,

I've been trying to run openBTS 2.4 for the last couple of days but
with limited success.

I use a Nokia 3310 in BTS Test mode locked on my ARFCN and it "sees"
the network but can't join. Basically it sends RACH (I need it's
monitoring AGCH on the lcd) but never gets any response.
Once, after a recompile I started OpenBTS just to see the results of
my additional debug and to my surspise, as soon as it started the
phone saw the network, sent a RACH and got assigned a SDCCH and
received the welcome sms and from there It worked, i sent sms thru
command line and such. But as soon as I stopped the process and
restarted it, never worked again ...

My main problem currently I think are RX timestamps discontinuity.

When looking at the  USRPDevice::readSamples function, I would expect
the timestamps on read packet to be continuous (i.e. lastPktTimestamp
+ 126 == newPktTimestamp). But in my case it's quite common to either
have overlaps (new packets has samples that were in the old as well
... didn't check if they matched) or big jump forwards (sometime tens
of thousands !!!). The transceiver process only takes like 20% of CPU
so I don't think that's the problem. I also tried setting it to very
high sched prio but that didn't help.

For the sake of full disclosure, I am trying to run it in a
'officially unsupported' configuration since I have only 1 RFX900
board so I'm doing RX/TX on the same board. I obviously had to change
the USRPDevice.cpp code to init the things properly. However I don't
think the problem lies there since during the tests, it worked nicely
for a few minutes, until I restarted.

Regards,

     Sylvain