Adaptive Control For Multimedia Stream In Desktop Video Conference System

The multimedia technology, computer technology and point-to-point networktechnology is important to the desktop video conference system. The network, theusers and the video conference system builds up an organic integer. It must meet therequirements of users by making use of the network resource effectively. At the sametime it must avoid the multimedia streams'impaction with congestion controlalgorithm. Mostly the multimedia applications emphasize the real-time capability,and usually adopt the UDP protocol to transmit multimedia streams. And the desktopvideo conference system terminals are different due to the different networkenvironment and the equipment. Thus the congestion control must be implementedon application layer. So, it is significant to research on the application layercongestion control algorithm of desktop video conference system and realize theadaptability for the network congestion.The model of desktop video conference system is the basement of congestioncontrol model. The paper first build a predigestion desktop video conference systembased on SIP protocol. Its concentrative control mode based on SIP is convenient forthe management in Internet. At the same time the distributed multimedia mode suitsthe multi-point to multi-point transportation. And the model is simple, open andextensible.Based on this simple video conference model, this paper gives MIMO adaptivecongestion control model. It based on adaptive control, is a point to point congestioncontrol model based on sending rate, and is aimed on the multi-point to multi-pointapplications. To reduce the flux of control channel and the response time, the modeladopted the UDP protocol to transfer the feedback info, and it didn't retransmit thelost message, just embedded a conservative fix module fixing the feedbackinformation for the validity of feedback information. To optimized allocatingbandwidth within the video conference system itself, the model used a clusteringalgorithm to classify the required bandwidths. When the CODEC output channelslower than the receivers, it can allocating the closer receiver to the same group. Thusfully utilize the limited bandwidth and optimize the receiving quality as a whole.The paper also implemented the algorithm based on VC++6.0, gave the artificeand some point should be pay attention to. For better performance, the paperdiscussed how to improve the operation speed, how to do the multi-thread scheduleand the NAT (network address translation) strike method. Finally, author established the experimental environment. And theexperimentation shows that when the congestion occurring, the algorithm is adaptive,its response time is shorter than existing algorithm. The experimentation also showsbetter TCP-friendliness.