A Self-Adapting Labelling Algorithm For Layered Multicast Based On Differentiated Services

With the rapid development of network multimedia technology, many emerging applications such as video on-demand, distance-learning and on-line game were introduced into Internet . Most of these applications require a large amount of network bandwidth, low delay and point to multipoint or multipoint to multipoint communicating capacity. Multicast technology is a very useful operation for supporting such applications, which can uses bandwidth effectively by sharing the links. Layered multicast can generate several multicast groups for layered video source such that the receivers can get video data with different quality by subscribing different groups according to their network conditions . So, layered multicast is one of useful mechanism for these new applications .On the other hand, the service model of best-effort is not enough good to satisfy the requirements of QoS that the new multimedia applications need. Differentiated Services (DiffServ) is a perfect model to provide scalable QoS . Multicast can reduce bandwidth consuming, and DiffServ can provide bandwidth allocation, they share complementary goals .But the integration of the two technologies brings new problems due to the fundamental conflicts of their architectures. The thesis surrounding how to solve these conflicts provided a solution.This thesis began with the basic concept of IP multicast, studied the multicast protocols, congestion controlling, layered multicast and layered video coding techniques .By comparing the merits and faults of DiffServ with IntServ, it expounded the function module of DiffServ and traced the process .It analyzed the advantage and problems while combining IP multicast with DiffServ .Than the thesis brought forth LMSLDS (Layered Multicast Self-Adapting Labelling Algorithm Based on DiffServ) . LMSLDS could make the forwarding nodes know the receivers' QoS requests clearly .It admitted members of group and allocated resource self adaptively .So LMSLDS can solve NRS and HMG problems efficiently .The simulation results of NS-2 demonstrated that this algorithm are a solution with better scalability.