[Asterisk-Dev] help needed implementing H324M (UMTS Video)

Klaus Darilion klaus.mailinglists at pernau.at
Sun Feb 6 14:36:34 MST 2005


Hi Christian!

Sorry for my late reply, but I was busy an the video project isn_t top 
prioritiy.

comments inline...

PS: I also cc to the mailing lists - maybe someone is interested in our 
ideas.

ckruetze at foni.net wrote:
> Hi,
> 
> I read your post on the * mailing List regarding your H324 implementation for
> Asterisk.
> 
> I'm very interested in that, and I would like to offer you my help.
> Unfortunately I'm somewhat new to * and therefore I can't answer your questions
> on how to access the raw bit stream from the channel.
> 
> Regarding where to implement it, I wouldn't modify the zaptel channel.
> zaptel cards are not the only way UMTS video could come it. There is also capi
> channel and maybe it could even come over SIP or IAX, who knows.

Yes, that's true. I think it would be good to write i as an application. 
The existing channels (zaptel, capi..) have to be modified to 
distinguish normal voice calls from video calls (bearer capability, user 
layer signaling) and handle the information the the dial plan. There I 
have extensions which match only videocalls and handle the call to the 
H324 application.
Then, the application handles the H324/SIP conversation and uses the 
internal codecs for transcoding (if necessary).

After digging a littl ebit more into the H.324 specification I realized 
that this gateway is much more complicated then I thought. The H.223 
framing uses the mobile levels which have much more complex frames than 
simple HDLC frames. You will have to code/decode Golay codes and have to 
guess the frames in case of bit errors. IMO this is not trivial.

After that, H.245 will be used for call control. Mybe its possible to 
reuse things from H.323, nevertheless I will take plenty of time to get 
familiar with the specifications. Another big problem is the asterisk 
architecture - it studied the code for a while and got an idea who 
things will work. Nevertheless, there is to less documentation - there 
are plenty of undocumented data structures and every time I used one I 
was hoping that I will use it in the proper ways.

Currently, we put the project "on hold" and hope to find some students 
which will implement it as diploma thesis.

If you are going to work on this project, I can help you with technical 
details about H.324M.

regards,
klaus

> I would try to implement H324M as a channel independent codec.
> First of all there are other ISDN channels other then zaptel, for example the
> capi channel or the i4l channel (can't remember it's name right now). It will
> also be more consistent with the current (voice only) codec’s and more flexible
> – not everybody wants that in there zaptel channel.
> 
> After asterisk "speaks" H324M, I would implement the translation into SIP video
> much like the different audio channels are currently translated from one to
> another.
> Of course for that to work, Asterisk would need to understand other codecs such
> as SIP video.
> Xten use in there eyebeam products H263 for video, so translation shouldn’t be
> too complicated.
> 
> Feel free to contact me by email, or MSN (same address as my email) or sometimes
> on the #asterisk chat (user: ckruetze)
> 
> 
> Looking forward to Video in Asterisk,
> 
> Christian
> 
> 
> 



More information about the asterisk-dev mailing list