Hi Adrian,
I am interested in the sdr_phy work being done currently
Great to hear that.
I built osmo-trx from the ms branch and implemented a crude power measurement command. When running trxcon everything works fine, I get power indications for channels, and then when the mobile app is trying to sync to an ARFCN I start having issues.
I have never developed / tested trxcon with OsmoTRX so far. Currently I am using GR-GSM as a L1 back-end, and the virtual connection between OsmocomBB and OsmoBTS during development.
In general, trxcon should be able to 'speak' with OsmoTRX, because one do use classical transceiver UDP interfaces. But there are some CTRL commands, which aren't implemented in OsmoTRX yet (such as ECHO and MEASURE).
I had to change the code of trxcon in order for SCH to work, because the bit values were inverted (0 was 1 and viceversa). I am attaching the changes at the end of the message.
Sounds strange. In case of virtual Um-interface (FakeTRX) and GR-GSM TRX all SCH messages are decoded correctly. I will look closer to this issue, thanks anyway!
It appears as if the deinterleaving in gsm0503_xcch_decode does not work... could be the order of bursts?
The order of bursts should be consistent, I mean frame-by-frame. You need to make sure that a frame number from SCH burst is being decoded correctly by OsmoTRX. Every burst coming from OsmoTRX has a header with frame number, which is used by trxcon to determine a current frame position within a multiframe. Probably, the problem is due to incorrect / consistent frame number values. The internal clock counter (sched_clck.c) is only used to schedule the UL bursts transmission.
I'd appreciate any help here. I'm also ready to share the changes I've made to osmo-trx/ms and trxcon.
For now, have a look at:
- GR-GSM based transceiver: https://github.com/axilirator/gr-gsm/tree/fixeria/trx
- Virtual Um-interface (brief description):
http://git.osmocom.org/osmocom-bb/tree/src/target/fake_trx/README?h=fixeria/...
- The latest version of trxcon (use fixeria/fake_trx branch): http://git.osmocom.org/osmocom-bb/log/?h=fixeria/fake_trx
You can also join the development process, feel free to mail me. I will create the wiki page as soon as possible.
With best regards, Vadim Yanitskiy.
Hi Vadim,
In general, trxcon should be able to 'speak' with OsmoTRX, because one do use classical transceiver UDP interfaces. But there are some CTRL commands, which aren't implemented in OsmoTRX yet (such as ECHO and MEASURE).
Yes, I implemented both commands, I can share the code with you. My problem is that Gnuradio (distro version) is built against an incompatible ABI version of libuhd, so I have to jump through major hoops just to get both osmo-trx and gnuradio working. So currently all my tests are using osmo-trx only.
The order of bursts should be consistent, I mean frame-by-frame. You need to make sure that a frame number from SCH burst is being decoded correctly by OsmoTRX. Every burst coming from OsmoTRX has a header with frame number, which is used by trxcon to determine a current frame position within a multiframe. Probably, the problem is due to incorrect / consistent frame number values. The internal clock counter (sched_clck.c) is only used to schedule the UL bursts transmission.
Yes, the SCH burst has a correct frame number. But the BCCH and CCCH bursts fail to be decoded. CRC fails in libosmocore/coding/gsm0503_coding.c Is your gr-gsm fake-trx able to decode correctly BCCH and CCCH? I double checked the osmo-trx code and besides some major differences between the MS branch and master, everything looks correct.
You can also join the development process, feel free to mail me. I will create the wiki page as soon as possible.
Yes, I'd like that, but for now I just want to get BCCH and CCCH working and the mobile syncing with the BTS. Then I will be able to test.
Cheers, Adrian
Hi,
Just a quick thing: to get ABI supported version of UHD do not use the Ettus ppa uhd, but use the one called uhd-lts or something similar. It is compiled with the right ABI version for the gnuradio distro package.
Cheers, Domi
2017. szept. 6. dátummal, 17:42 időpontban Adrian Musceac kantooon@gmail.com írta:
Hi Vadim,
In general, trxcon should be able to 'speak' with OsmoTRX, because one do use classical transceiver UDP interfaces. But there are some CTRL commands, which aren't implemented in OsmoTRX yet (such as ECHO and MEASURE).
Yes, I implemented both commands, I can share the code with you. My problem is that Gnuradio (distro version) is built against an incompatible ABI version of libuhd, so I have to jump through major hoops just to get both osmo-trx and gnuradio working. So currently all my tests are using osmo-trx only.
The order of bursts should be consistent, I mean frame-by-frame. You need to make sure that a frame number from SCH burst is being decoded correctly by OsmoTRX. Every burst coming from OsmoTRX has a header with frame number, which is used by trxcon to determine a current frame position within a multiframe. Probably, the problem is due to incorrect / consistent frame number values. The internal clock counter (sched_clck.c) is only used to schedule the UL bursts transmission.
Yes, the SCH burst has a correct frame number. But the BCCH and CCCH bursts fail to be decoded. CRC fails in libosmocore/coding/gsm0503_coding.c Is your gr-gsm fake-trx able to decode correctly BCCH and CCCH? I double checked the osmo-trx code and besides some major differences between the MS branch and master, everything looks correct.
You can also join the development process, feel free to mail me. I will create the wiki page as soon as possible.
Yes, I'd like that, but for now I just want to get BCCH and CCCH working and the mobile syncing with the BTS. Then I will be able to test.
Cheers, Adrian
Yes, I implemented both commands, I can share the code with you.
It would be great. Just send these patches here using git format-patch / send-email. I'll have a look.
My problem is that Gnuradio (distro version) is built against an incompatible ABI version of libuhd, so I have to jump through major hoops just to get both osmo-trx and gnuradio working. So currently all my tests are using osmo-trx only.
Ah, just build it yourself. It would be much better. Distro versions are often pretty dated :( Excluding Arch Linux and similar ;)
Yes, the SCH burst has a correct frame number. But the BCCH and CCCH bursts fail to be decoded. CRC fails in libosmocore/coding/gsm0503_coding.c
Regarding to the 'bits conversation' you mentioned in the first mail, it's done the same way as in OsmoBTS. It should work as is. Probably SCH bursts are not actually forwarded from OsmoTRX. Or being forwarded in a wrong way.
Is your gr-gsm fake-trx able to decode correctly BCCH and CCCH?
Sure. I am even able to make a voice call.
With best regards, Vadim Yanitskiy.
baseband-devel@lists.osmocom.org
-
Adrian Musceac -
Tomcsányi, Domonkos -
Vadim Yanitskiy