Hello Osmocom community,
I am pleased to announce that my new E1 Abis MGW (replacement for the limited E1 Abis function in osmo-mgw) now supports AMR speech codec, in addition to FR/HR/EFR and CSD:
https://gitea.osmocom.org/themwi/tw-e1abis-mgw
Basically compile it from source following instructions in the README file, run it alongside with all of the standard Osmocom components, and enjoy the addition of AMR and CSD capabilities - if you have E1 BTS hardware, that is...
(For those not familiar with E1 BTS support in Osmocom, the E1 Abis MGW function of osmo-mgw supports FR/HR/EFR, but not AMR or CSD.)
I realize that this news item may not be of much interest to most people nowadays, simply because almost no one uses the combination of Osmocom CNI stack with E1 BTS hardware any more. However, there is little in the way of hardware alternatives: if someone wishes to operate a traditional high-power, high-capacity GSM cell site (typical 3-sector geometry, high Tx power, multiple TRXs per sector for decent capacity), I am not aware of any publicly available hardware that can achieve this job in the "modern" way of SDR etc. Hence finding old Ericsson or Nokia BTS hardware on the surplus market, just old enough to still offer E1 interface as an option, is the only practically available choice for a professional-quality cell site with Osmocom software stack from the BSC up.
And why does it have to be E1, why not more "modern" hw from the same Ericsson/Nokia/etc that still supports GSM, but only with IP interfaces? Answer: those IP interfaces are proprietary, and we have no docs - or at least I don't. OTOH, E1 is highly standardized, and for those parts that differ per vendor, our little community already has the necessary knowledge...
I will now be preparing some patches for osmo-bsc and its underlying libosmo-mgcp-client to support switching on TW-TS-006 and TW-TS-007 RTP enhancements in the E1 Abis MGW when the CN requested them via BSSMAP extension of TW-TS-003, and then I will need to get back to work on CN software, namely a more proper MSC. I now have a new modular architecture in mind, in which osmo-msc will be just one component of what I consider to be the MSC as a whole. osmo-msc will need some patches in order to work with a different MGW (one that transcodes to G.711) and some MNCC changes, but I'll implement them in such way that all functional changes will be enabled either by vty config or by bits set in MNCC struct by external MNCC sw, so those patches will stand a chance of being mergeable. But the part with osmo-msc patches is still far away...
GSM/2G Forever, Mother Mychaela