An Improved Algorithm Of NICE: Based On Topology-Aware And Capacity

Since 21st century, the rapid population of the Internet and the upgrade of network bandwidth make it possible to obtaining a variety of services that demands high bandwidth. Such programs using unicast technology taking the C/S model provide a single data transmission channel from special servers to every user. It’s possible for a small user base, but with the rapid growth, network bandwidth has to greatly enhance, and the enterprise cost has to greatly increase too. International scholars proposed multicast scheme. IP multicast was first proposed, and working at network layer. IP multicast has a lot of disadvantages, such as accounting difficulties, excessive consumption of router resources, etc. These disadvantages have made the IP multicast cannot be deployed on a large scale, which greatly restrict the development of this kind of application. In recent years, a new kind multicast scheme was proposed to replace IP multicast, which is the application layer multicast based on "End-to-End Argument" and working at application layer. Application layer multicast use unicast to archive multicast, and data is copied and forwarded by host. In order to obtain low delay, good performance and stable service, high efficiency and stable application layer multicast tree should be build. So how to build a low delay and stable application tree is always the research focus in the field of application layer multicast.The NICE protocol proposed by Banerjee Suman in 2002 is one of the most widely studied application multicast protocols. NICE protocol adopts the idea of layering and clustering, solved the problem of constructing large scale multicast tree with only one data source. NICE protocol has the following two main problems:1)many neighbor nodes cannot be clustered, and this situation is increased by frequent cluster split; 2) the selection of leader of cluster and nodes’father just care about delay, the uncertainty and differences in capacity of terminal equipment is seriously affect the stability of multicast tree. So In this paper, we mainly study the two disadvantages and propose an improved algorithm.Firstly, we analyze the recent research progress and research results of application multicast in the world and finally made a summary.Secondly, we proposed a topology-aware partition clustering model (called TPCM), TPCM can just split the foreign nodes, and localize the transmission of data. Furthermore, we proposed an improved algorithm based on TPCM. Finally the effectiveness of the improved algorithm was verified by simulation experiment.Finally, we proposed a new strategy based on nodes’capacity to improve the NICE protocol. Delay, online time and bandwidth are all considered in this new strategy called BONC-NICE (Based on Node Capacity), so this protocol can better simulate the network situation in the real world and can improve the performance and stability of multicast tree. We also verified the strategy by simulation.