The initial idea was to give mobile users the ability to broadcast live from their mobile phone to the internet. The jury´s remarks extended the topic to video streaming.
Live streaming offers a new type of media to the Web 2.0 idea of user generated content. Integration in social networks (Twitter, Facebook) leverages the application by using existing communities.
A lot of trial and error happened during the sprint, resulting in a lot of non-working approaches, which taught us a lot about SIP architecture, challenges working in the mobile handset environment and problems regarding current video SIP implementations.
It took some time to realize that video streaming into an IP base environment is currently only barely supported on most handsets.
An approach trying to use J2ME to create an environment running on most phones failed because J2ME just doesn't support (yet) streaming live video from a mobile handset.
After some discussion with tech staff from A1 it became clear that streaming an UMTS video call to SIP isn't possible at the moment as well, as A1 is still working on the technical difficulties of this solution.
So we settled for using the Movino client on the handset, which uses an M-JPEG stream embedded into a proprietary stream format, so bypassing most of the restrictions currently present in the mobile environment. The only issue with this is that the energy consummation is very high on the handset, and in a high-resolution setting even with the AC adapter present at the phone (Nokia N95), it still needed to use battery power to supply the camera and coding of the live stream. So some solution using the internal chipsets of the mobile would be highly appreciated. The Movino design description contains some thoughts about this topic.
The Movino client is at the moment mainly supported on the Symbian S60 platform. The Movino J2ME client sends no audio stream and up to five snapshots per second.
During the sprint it became apparent that the SIP video support in the open source SIP stacks is still rudimentary. The only currently-known (by us) SIP stack which supports some kind of video on a higher level language (without resorting to program in C) is the MJSIP stack for Java. Test using the pjsip stack for python failed, apart from having massive build problems.
The overlap between the codecs available in the Open Source SIP clients and the commercial SIP softphones is very small, sometimes even not existing. This aspect made especially the testing hard.
The discussions with the other teams and the support staff brought up new ideas:
What happened? What has been created? What successes have been reached? What would you like to tell others (to avoid double work)? What new ideas did emerge? What have you learned?
This is the archicture how it finally has evolved during the prototyping sprint, and how it would be implemented towards a production release.
Some of the components have already been implemented as a proof-of-concept (e.g. flash streaming, audio streaming to SIP clients)
At a party, there could be devices like monitors or video walls which show user-streamed content. The party crowd could stream live videos from parts of the party to those videos walls, interactively helping to pattern the party. The vidoe content would also be streamed to the Internet, so people not able to attend to the party would also get impressions from there.
Inform your friends with small video clips about what you are doing at the moment. This is like Twitter or SMS without the need to type something.
It would be possible to play something like „Scotland Yard“ (TM) in real time. The hunted one has to broadcast every public-transport-ticket and his POV every 15 minutes using the video streaming, and the other ones in the game need to circle and catch him.
Many open source Flash video player are available, e. g.
The PHP application deliverd with Movino provides ready-to-use integration to:
The prototype was realized using OS FLV.