A Research On Aggregated Models And Algorithms Analysis Of Multicast Forwarding States

IP multicast is an efficient multipoint-to-multipoint delivery mechanism in IP network, and accelerates the development of multi-point communications services such as Video on Demand, video/audio conferences and data distribution. While IPv6 strengthen its support on multicast and multicast tech. gets a large advantage over unicast and broadcast. Recently, multicast is used more and more widely, the demand of network is more and more intense, and the huge superiority of multicast compare with unicast or broadcast, people begin realize the advantage and benefit of multicast.However, a multicast distribution tree requires all the tree nodes to maintain per-group forwarding state, so when there are large numbers of multicast groups in the network, IP multicast suffers from many problems: The number of forwarding states obviously increases with the growth of the number of groups. The increasing growth of the forwarding states requires the growth of memory and CPU process of the router, and the forwarding process will slow down. Therefore, when there are a large number of simultaneous multicast sessions, it will cost much resource and control spending on the groups' management of multicast, in other words, the state scalability problem has restricted the application of scalable multicast.To solve the problem of scalability, a novel approach called aggregated multicast has been proposed, which is based on the feature of actual networks. It's first proposed by the UCLA network laboratory, and it's followed with the Greedy Algorithm. Its main idea is to broaden the demands of saving bandwidth properly, and to enable many multicast groups to share a multicast distributing tree. In this way, the number of multicast trees is decreased deeply, and accordingly the forwarding states of multicast is decreased, and finally the performance of network is improved.In this paper, two different algorithms are proposed to solve the forwarding states problem of multicast after achieving extensive and intensive research on the area of traditional aggregated model of multicast: a scheme based-on Tree Superposed algorithm and a scheme based-on Genetic algorithm.1. The Tree Superposed Algorithm borrows idea from the concept of "similar" of graph theory. Making original multicast trees superposed according given special demands of QoS, and then pruning and cutting rings and the like by the definition of tree, and finally get the set of aggregated multicast.The results of simulation indicate that the superposed tree algorithms are superior to the conventional greedy algorithm on the time performance, aggregated degree and forwarding state reducing.2. In Schemes based on Genetic Algorithm and Simulated Annealing Algorithm. Simulating annealing algorithm can get the hypo-global optimal solution by enough time, while Genetic Idea emphasizes on the evolution relations between two populations, but the mating operation may lost the best solution. Therefore, in this paper, we provide an approach based on Genetic and simulated annealing algorithm.The results of simulation indicate that the superior performance of Genetic Algorithm. By comparing to the Greedy Algorithm and Lagrange Algorithm, the Genetic Algorithm get better results on the aspect of both aggregated degree and forwarding states reducing.