Research On Scalable Multicast Scheme

With the rapid development of Internet transmission and processing capacity, the multimediaapplications have become the fundamental service of Internet. Compared with unicast andbroadcast, IP multicast, as a group communication method, could improve efficiency of networkresources usage, and provide better transimission service for multimedia applications. However,complex routing scheme and high maintenance cost reduce the scalability of IP multicast, andmake it hard to be implemented in real network. With the fundamental of National973Program“Flexible Architecture of Reconfigurable Infrastructure (FARI)”, this thesis focuses onimproving scalability of IP multicast in four aspects:1. Aiming at improving convergence speed and reducing routing cost in multiple QoSconstrains, an improved Harmony Search algorithm based on Child-Node Encoding (CNE-HS)is proposed. In this algorithm, a new encoding mechanism based on child node and a newdynamic tuning method for parameters are proposed, which help to accelerate iterative speed andreduce multicast routing cost. Simulation results show that CNE-HS could achieve24.2%costreduction and24.3%iteration decrease compared with original harmony search algorithm, whichindicate that CNE-HS could improve the performance of routing algorithm efficiently. Byexpanding the scale of a single multicast group, CNE-HS improves IP multicast scalability.2. Aiming at reducing maintenance cost of forwarding states for large numbers ofmulticast groups, an Aggregated Multicast algorithm based on Link Redundant Degree(LRD-AM) is proposed. By giving the definition of link redundant degree, LRD-AM could judgethe similarity of different multicast trees. Based on tree similarity, LRD-AM makes multiplegroups aggregated to share one delivery tree, which reduces multicast forwarding states.Simulation result shows that LRD-AM could achieve65%forwarding states reduction with3000multicast groups, which indicates that LRD-AM could reduce the maintenance cost offorwarding states efficiently. By expanding the scale of coexist multicast groups, LRD-AMimproves IP multicast scalability.3. Aiming at balancing resource usage when multicast routing change dynamically, aReconfiguration Algorithm based on Resource Effect Degree (RED-RA) is proposed. In order todescribe the importance of different resource, RED-RA gives the definition of resource effectdegree (RED). Using RED, cost function of path construction is proposed. For three differentscenes: application requirement changing, multicast member changing and network changing,RED-RA accomplishes fast dynamic reconfiguration of multicast routing and resource usageoptimization. Simulation result shows that, compared with other algorithms, RED-RA couldachieve25%reconfiguration time reduction and15%cost reduction, while getting betterresource balance. By optimizing network resource utilization, RED-RA improves IP multicastscalability.4. Based on the three above algorithms, a Scalable Multicast Scheme supportingReconfiguration (SMSR) is proposed. Based on cognition of resource and applications, SMSRcould accomplish construction, aggregation and reconfiguration for multicast routing. To validate the feasibility of SMSR, reasonable implementation scheme, test scene and scheme are proposed.Test results show that SMSR could accomplish the schedule functions as expected, whichprovides a new idea for solving IP multicast scalability problem.