Hi Harald and list,
Which model of available femtocells are you focusing on and is it easy
to purchase one (on ebay perhaps)? Cause for 2G the ip.access's nanoBTS
imho is best supported by openbsc. If you can give us some more
information what device to use (as a start), than we can at least focus
on one device and extend to other models in the future.
On 16-12-2010 10:09, Harald Welte wrote:
Hi all,
i just wanted to keep you posted with the 2011 roadmap for 3G integration
into our "Free Software Mobile Network" stack. This message summarizes
the plan of what I expect to be working on. Comments welcome.
3G support directly in OpenBSC is, however, a bit more difficult than we
originally anticipated. The primary problems are:
* 3G and 2G share the same core network (CN = MSC/VLR/HLR/AuC/SGSN/GGSN)
but have different radio access netowrk (RAN = BTS+BSC vs. NodeB+RNC)
* the interface between RAN and CN is different, i.e. 'A' interface
(08.08) in GSM case, RANAP in 3G case.
* The encoding of the two protocols could not be any different. Hand-coded
vs. complex ASN.1 structures
So with OpenBSC, we face the following challenges:
1) We implement the GSM-RAN functionality and the CN in one process,
without any separation or standardized interfaces between them
2) Implementing RANAP in C can be done, but is probably quite cumbersome
given the tons of code asn1c generates and how difficult it seems to
get all the memory allocation/freeing right in all the cases
What we instead decided to do is:
a) Split the BSC part from the CN by implementing the A interface (this
is already finished, we now have osmo-bsc to do exactly that)
b) Implement a new MSC part, using the A interface towards the BSC
This is still possible using our existing code in the original
OpenBSC, though we have not completed/finished it yet.
The real goal is to have a new MSC, written in a higher-order language
that provides proper tools for working with ASN.1 - e.g. Erlang. I have
just started working on this and am more or less finished with the lower
layers like SCCP and am moving into 08.08 and the actual layer3 protocols.
There is already some initial funding secured for the first months of
this work, and funding will likely continue.
c) Once the 2G functionality of the Erlang MSC is working, add the RANAP
interface to it.
At this point we should more or less be able to directly
connect the Alcatel-Lucent femtocells, as they basically consist of
HomeNodeB plus an internal RNC handling the RRC protocol.
d) Adding support for ip.access/Cisco/AT&T will need an implementation of
their URSL protocol, which is proprietary. URSL can either be added
to the new MSC directly, or a URSL<->RANAP translater be developed.
e) Adding support for the Ubiquisys/SFR femtocell would require us to implement
the UMA protocol stack. This can once again be either done as part of
the MSC or using an external translator/gateway program.
Regards,
Harald