Research On Application Layer Multicast And Steiner Algorithms

With the improvement of the network bandwidth and handling ability, network provides us with more multimedia services and many of these services require network with multicast capacity. IP multicast is the first effective multicast, but because of technical problems and marketing problems caused by IP multicast, it has not applied successfully in the global scope. Even in several years when IP multicast was put forward, there still were not a successful IP multicast deployment in the internet. So the solutions of application layer multicast were put forward.The basic idea is to achieve multicast services in the application layer for users without any change in the network infrastructure and without relying on multicast services provided by the network layer. At present, there are many methods to solve application layer multicast problems in this article. Although Steiner tree problem is a historic mathematical problem, it has extensive application in network design, the multicast routing, and VLSI design aspects. Considering the corresponding relationship between multicast and Steiner, this paper focuses on Steiner heuristic algorithms.In this paper, we introduce multicast technology and summarize the application layer multicast with distinguishing the characteristics between IP multicast and application layer multicast. Heuristic algorithms based on Steiner are the currently effective methods of solving multicast. This paper gives the definition of Steiner, and introduces the exact solution algorithms, typical heuristic algorithms and greedy algorithms briefly. Then it summarizes a framework of typical heuristic algorithms. Based on this framework, designing Steiner heuristic algorithms based on the key nodes and the key edges of the path, which make the added paths reuse the key edges and nodes, and analysis the correctness and the superiority of the algorithm through examples and experiments.Lastly, the paper designs improved Steiner tree heuristic algorithm based on weight, which combines the important information of the key nodes and key edges based on basis heuristic algorithms. The experiments show that the method with less computing time and higher quality.