Researches On QoS-Aware Multicast Strategies

Over the latest several years, the phenomenal growths of users'quantity and constantproducing of new applications have accelerated the development of new Internet technologies.The dominant question facing the Internet today is, how can the network meet the needs of theusers and their applications while trying to keep such implementations scalable to the billions ofusers present on the Internet? Two of the emerging technologies for answering the question areDifferentiated Services (DiffServ) and multicast. Multicast can save bandwidth while DiffServhas been proposed as a scalable solution for supporting QoS in the Internet, especially, it islikely that some form of DiffServ model will be implemented in the next generation Internet.This thesis emphasizes the combination of IP Multicast and DiffServ so as to provide anefficient DiffServ-Multicast strategy.Coexistence of multicast and DiffServ is promising since the DiffServ model can provide ascalable framework and may reduce the computational complexity to locate a QoS-satisfiedmulticast tree. Although the two technologies share complementary goals, the integration of thetwo technologies is a non-trivial issue due to three fundamental architectural conflicts, simplecombination of the two would bring about such problems as Neglected Reservation Sub-treeProblem (NRS), Dynamics of Arbitrary Sender Change and heterogeneous QoS requirement ofreceivers. These problems mainly lead to the situations in which other service users are affectedadversely.In this thesis, the available DiffServ-Multicast solutions are classified into State-based,Edge-based and Encapsulation-based types, meanwhile the advantages and disadvantages ofeach type are detailed.After analyzing the issues essentially, we propose an efficient Encapsulation-Based scheme--QoS-Aware Multicast over DiffServ Domains (QoSAMDD), which solves the problem ofextensive group size, number of groups, heterogeneous QoS supporting and dynamicmembership join/leave. By this way, the quality of the Multicast traffic can be assured. Firstly,the basic transport mechanism, tree structure encapsulation field (TSEF) and variable QoSExtensions are introduced in details. And then, theoretical analysis has been done on the cost ofthe state in the core, the overhead versus the other two approaches, the effect of fragmentationand timer T. Also, a great deal of experiments have been done on effect of packet size, effect ofgroup size, heterogeneous QoS supporting and the share with TCP with NS2. According to theTheoretical studies and experimental results. Despite the fact that QoSAMDD incurs per-packetoverhead due to the TSEF, it can offer the performance while dramatically reducing the state...