Video delivery in packet switched networks like the Internet is a challenging issue. Internet provides a best-effort service, meaning that there is no quality of service guarantee in the current Internet, and network parameters like the bandwidth, delay and packet loss probability vary dynamically in the network. Besides, other issues like optimal video multicasting and fare bandwidth allocation between video streams and other traffics in the network remain to be solved. In this thesis we first review the concepts of Multiple Description Coding and Rateless Coding and then propose a new method for multicasting video on IP-based networks. In the proposed method we used both multiple description coding and rateless coding to make video streams more resilient against packet loss in the network, and in addition provide different quality levels for heterogeneous users. In this method we used even-odd frames separation to create two descriptions from the original video and Raptor codng is used for the rateless coding part. For multicasting the video on the network, we proposed an algorithm for receiving feedbacks from the users to control the amount of overhead sent by the server according to the packet loss probability. This algorithm keeps the number of feedback messages minimum in order to prevent congestion in the server side or the network. Finally the simulation results are provided and the proposed method is compared to simple multicasting of the video. The simulation results indicate a great improvement in the users’ video quality. KEYWORDS: Video Multicasting, Multiple Description Coding, Rateless Coding
