Hi,
I further implemented the base of the virtual layer with the purpose
of replacing the um-air interface with a multicast-socket solution.
This has been started by Harald Welte some time ago in the Osmobts -
laforge/virt-bts branch.
As I was not really confident enough to create an official new branch
before someone had a look on what I did, so I developed in copies of
the osmocom-bb and osmo-bts projects on my github-page
(
https://github.com/BastusIII/osmocom-bb-virt/tree/stumpf/virt-um and
https://github.com/BastusIII/osmo-bts-virt/tree/stumpf/virt-bts).
If someone finds to time to have a look into them (structure, path,
implementation) and give me feedback about it, I would be glad :).
Please use the following config files that i have tested:
-
https://github.com/BastusIII/osmocom-config-files/blob/master/openbsc-virtu…
-
https://github.com/BastusIII/osmocom-config-files/blob/master/osmobts-virtu…
-
https://github.com/BastusIII/osmocom-config-files/blob/master/osmocom-bb-mo…
What works:
BTS:
- downlink over gsmtap and a multicast socket
- OML and RSL on abis properly established and seems to work
- virtual-um connection establishment
Osmocom-bb:
- virtual-um -- l23 app connection establishment (bridging osmocon,
no serial link needed)
- BCCH downlink handling and forwarding to l23 app
- handling of some l1ctl requests from l23 (e.g. power management had
to be mocked, so l23 gets a response and is satisfied)
What not:
BTS:
- missing uplink handling
Osmocom-bb
- missing uplink routines (RACH, TCH, dedicated channels)
- handlers for l23 requests only partially implemented (missing TCH, RACH, ...
I a currently a bit overwhelmed by the mass of messages exchanged
between l23-app (used mobile, btw.) and the virtual-um and hanging
because mobile gets a sync timeout.
I am looking forward to hear from you :).
Sebastian Stumpf