Harald Welte's blog


Harald's Web




Other Bloggers
David Burgess
Dieter Spaar
Michael Lauer
Stefan Schmidt
Rusty Russell
David Miller
Martin Pool
Jeremy Kerr
Tim Pritlove (German)
fukami (German)
fefe (German)
Bradley M. Kuhn
Lawrence Lessig
Kalyan Varma


Ohloh profile for laforge
Linked in

Creative Commons License
Articles on this blog/journal are licensed under a Creative Commons Attribution-NoDerivs 2.5 License.



Sat, 06 Jun 2009
OpenBSC on its way to get funded A-interface development

There is more commercial interest in OpenBSC than I expected initially when I started the project as a 'just for fun playground for GSM hacking'. Now we have received an inquiry from a company who wants to fund the development of an actual A interface to OpenBSC. This basically means that somebody wants to hook up OpenBSC to an actual real-world MSC (Mobile Switching Center) of an actual real-world GSM network.

The A interface is the interface between the BSS (BSC + BTS) and the higher levels of the telephony network. The interface is based on SS7 and lives on top of SCCP. There's BSSMAP, DTAP and OMAP. Both connection-less and connection- oriented modes of SCCP are required.

What this means is that OpenBSC's software architecture will shift even further towards the traditional GSM network architecture. So far, we have a "full GSM network from BSC to MSC/HLR in a box" approach. This makes it easy to implement, but is quite restrictive. You cannot route/switch calls to a different network, e.g.

The recent patches posted by Andreas Eversberg already introduce a software interface called mncc into the OpenBSC codebase. While those patches are not merged yet, they are introducing a functional split between the call-control entity on one hand, and the RR and NM as well as Abis RSL/OML functionality on the other side.

When we introduce the A interface, the functional split in the software will be driven even further. We'll first introduce an API at the traditional BSC/MSC split, and then write a BSSMAP/DTAP/OMAP protocol interface to that API.

One thing is for sure: We'll always keep the 'run everything in one box' mode around. This is still the most useful case for small-scale experimentation with GSM.

I'm definitely looking forward to see this project grow. We still have no agenda for things like GPRS/EDGE support, or any kind of handover. But then, development one step at a time is more healthy than trying to do everything at the same time.

I'm really excited to play with the A interface, and to interact with an actual MSC on a protocol level. This sort-of completes my ventures into GSM protocol land, from the Um (on-air) over A-bis to the A interface, one iteration up the network hierarchy at a time.

[ /gsm | permanent link ]

On my way to FreedomHEC Taipei 2009

In about 8 hours I'll depart for FreedomHEC Taipei 2009, an event where members of the Linux development community try to help Taiwanese hardware vendors understand the Linux development model.

I personally believe this kind of event could not be any more important. The traditional PC and embedded hardware industry still has a very, very limited understanding when it comes to properly supporting Linux, aiming at the universal solution for best end-user experience. In order to achieve this, the FOSS development model needs to be understood, as well as the value of going mainline with the drivers/ports.

Once that point is reached, there needs to be understanding _how_ to achieve that. Availability of documentation is another key issue. If you want to enable people to help you with development, bug fixing and maintenance, you need to release programming manuals for the hardware..

I'm happy to see that this year the organizers were able to get prominent speakers such as Jon Corbet from lwn.net, and Greg K-H who is doing marvelous work with his Linux Driver Project. Last, but not least, Peter Stuge will be presenting on coreboot as a FOSS alternative to legacy BIOS.

I'm also happy to see more native/local speakers, such as the presentations by jserv (aka Jim Huang) on Qi, the bootloader that was developed as part of Openmoko - or the presentation on VIA's experience of merging code mainline by Joseph Chan.

[ /linux/conferences | permanent link ]