Extending VideoLan to support Distributed Streaming

Hallo,

we are a team of 4 postgraduate students attending the Master's in
Information Networking course offered by Carnegie Mellon University(
www.ini.cmu.edu). In the framework of a course in Distributed Systems we
have undertaken a semester project on Distributed Video Streaming over the
Internet. The idea we had is to provide a fault tolerant video streaming
service based on the principles of replication and redundancy we are
examining in the course. As we see it and after some short research in the
Internet for similar activities we have found out that there are 2
approaches to achieve the above.

a. Have multiple sources, whereby only one source transmits at a time and
in the case of a network failure another source takes over from the exact
same point that the previous source stopped. A good synchronization
protocol between the servers or perhaps some feedback mechanism on the
side of the client combined with some adequate buffer support on the
latter should make this "handover" seamless to the viewer.

b. Multiple sources transmit at the same time contributing in an additive
way to the quality of the final stream. If one source goes down for
whichever reason the quality will be reduced but will remain sufficiently
good for the user. This approach can make use of the layered architecture
of one of the encoding formats such as MPEG-2 and can even be designed in
such a way that will support some kind of prioritization! For example in
the case network congestion the server might select to  gradually drop
some audio channels, subtitles, some P frames etc.

In any case we are not experts on the field and we don't know much about
either MPEG-2 or streaming. Of course we definately don't aim in developing
streaming applications or video codecs from scratch. We have found the
VideoLan Software on the Net and thought that with some reasonable amount
of effort we could hack some "distributed streaming" functionality into
it. For this reason there are a couple of things that we would like to ask
some experts out there. If you are one of the developers of VideoLan and
you find interesting our proposal we hope that you keep on reading and
send us an e-mail with some answers/comments on the following.

1. Is there any activity from the VideoLan Community to support the above
described "distributed streaming" in any form?

2. Which would you think would be the most interesting of the 2 approaches
mentioned above and which the most difficult to implement?

3. Is the VideoLan Community insterested in such an extension of the
Software and if yes would you be willing to provide us with some support in
our effort to implement it. For example we could definately use the opinion
of someone of the Software's developers on where should we make changes to
implement the communication between the servers etc.

4. Which video format would you consider more appropriate and more feasible
to use for the above mentioned purposes.

So these are our basic concerns for the moment. I hope the size of this
e-mail didn't discourage any interested people to read it till the end. If
someone wants to get in contact with us we have just launched a site at
http://members.lycos.co.uk/insomniacsgr/ with some initial information
about who we are and what we do.

If you've made it this far thanks very much for the patience and we hope to
hear from you.

                            On behalf of the "Insomniacs"

                                    Panos Kassianidis


-- 
This is the vls-devel mailing-list, see http://www.videolan.org/streaming/
To unsubscribe, please read http://developers.videolan.org/lists.html
If you are in trouble, please contact <[email protected]>