From: David A. B. <dbu...@jc...> - 2009-12-13 02:17:14
|
Erik - On Dec 12, 2009, at 5:09 PM, meetmecall wrote: > I have tried to get a phone registered for days, making little > adjustments and following advice. I received SIM cards that has > proofed to work with OpenBTS and again no results. I had my my USRP on > a desk with some Epia ITX motherboards running and some other devices. > I moved the OpenBTS setup to the living room and the miracle happened: > I had a registered Nokia 3330. I could make a phonecall to a Music On > Hold extension I set up in Asterisk, I could send an SMS from the > OpenBTS cli to the phone and with a call file in Asterisk I had the > phone called by the Asterisk server. Is it really possible that a > couple of Epia ITX boards can make a USRP not work properly or is this > just a strange coincidence. It may be possible. We've seen problems with some older Dell laptops coupling a lot of RF noise into the USRP over the USB cable, so, yes, it is possible that EMI from some hardware is breaking the system. Which SIMs do you have? China Mobile? Tesco? If so, that's good because those carriers don't have coverage or roaming agreements in NL and the SIM is not the problem. But if you are using a SIM from any major carrier who has coverage or even has roaming agreements in your country, and you are not controlling network preferences in the handset, then you will have a serious problem controlling your test. Even an expired SIM will strongly prefer its home network and will camp to it quickly and ignore your system. When that happens, that is a case of operator error, not a failure of OpenBTS. > > When the phone looses registration because I walk it out of range it > doesn'r reregister when the phone is moved within the range of > operation of the OpenBTS GSM access point again. If I restart the > server and the usrp and when I completely restart the Nokia 3330 after > a while (5 minutes 10 minutes) the phone is registring again. It seems > to work in a way but it is very unpredictable, the registration > process takes a long time and reregistring isn't working at all (at > least not within 15 minutes). Is there anything in the settings I can > change to make it work better and make OpenBTS more aggressive in > discovering and registring new phones. Have you actually looked at the GSM spectrum in your area? I know the urban European cellular bands are crowded. Are you sure that you are not in direct interference with a carrier in your area? > > I tried to send an sms to the phone itself and used the IMSI number > for it. This makes the OpenBTS hang with this output over and over > again: > > fusb: short write xfr: 0 != 2560 Yes. This is a known bug, not in OpenBTS itself but in the USB interface used by OpenBTS. It happens on some systems, not on others. It is fatal on some systems, not on others. For example, I never see this under Mac OS X. It may be possible to prevent this error through a correct BIOS configuration, but this bug really needs a lot more research. See http://sourceforge.net/apps/trac/openbts/ticket/64. Anyone is welcome to add some notes to that. The long-term solution is probably to go away from USB all together. > > Is this a known bug or should I rapport this to a bug tracker .How am > I suppose to send an SMS from a phone to a phone. Sending an sms from > the OpenBTS cli works fine. > > Now and then I have a message in the OpenBTS cli: > > ortp-warning-Must catchup 57 milliseconds > > Is this a serious warning if it shows op a couple of times a day. I > installed ortp from ortp-0.15.0.tar.gz That message usually means that the CPU failed to keep up with real time, often as the result of a swap operation in some other process. The system should recover perfectly, though. > > > Thanks in advance. > > > \erik > gnuradio 3.2.2 > OpenBTS 2.5 > otrp > David A. Burgess Kestrel Signal Processing, Inc. |