Scalable video streaming in wireless mesh networks.

Wireless mesh network provides efficient and reliable services for large scale communications. Video streaming in wireless networks enhances the services by delivering multimedia information to end users. However, because of the dynamic conditions of networks and variety of users, how to smoothly deliver the multimedia data to users without wasting precious network resources is still a challenge. This thesis addressed this challenge by investigating several key issues in video streaming in wireless mesh networks. Firstly, a video streaming system, Swan Video Streaming system (SVS), over wireless mesh networks was designed and developed. Secondly, a scalable video coding scheme was adopted in SVS. Video bit streams were split into two layers, base layer and enhancement layer. These two layers of video streams were packed into two multicast groups to allow users to get access them separately based on their processing ability and network conditions. This prevents the waste of network bandwidth by eliminating the delivery of videos to all the users regardless of their conditions. Thirdly, to improve the video robustness and reduce the overhead of the network for real-time video streaming, the important parameter messages of scale coded videos are transmitted in a reliable manner. SDP (Session Description Protocol) and RTCP (Real-time Transport Control Protocol) were improved to transmit the control messages at the beginning of video transmission and during video transmission stages, respectively. A new rearrangement method in RTCP of received packets was also proposed to improve the efficiency of algorithm and reduce network overhead. In addition, based on the feedback from video server and receivers, server and receivers can adjust their output bit rate and receiving rate according to different conditions of network to reduce the congestion. The above approaches have been evaluated in the developed SVS testbed. Tests results show the approaches are effective and feasible in real application scenarios