Finding the highest performance for one-to-many content distribution in peer-to-peer networks based on the clustering and loading bandwidth of peer Nodes
Abstract
Considering the problem of IP multicast implementation in routers, in recent years, a lot of alternative methods have been introduced of the application layer multicast (ALM), for one-to-many content distribution. The present study aims to provide a new Algorithm in the field of ALM, to reduce the delay in peer to peer content distribution network (P2P), based on cooperation of M-ary and cluster nodes. All nodes which are close to one another gather in a cluster by means of a fixed number of landmarks that are known nodes. After the close nodes come to each other in a cluster, a tree structure is used to connect them. The algorithm is based on a cooperation between the source node and the content requesting nodes. In this algorithm, the source divides the content into blocks and the blocks are distributed in each cluster through m-ary trees that are all rooted in that source. Based on the mechanism used in this algorithm, all the participating nodes are used as a distributor of content, at least for one time. This algorithm exploits maximum upload capacity of the participating nodes and maximizes the final throughput. Due to the proximity of the nodes in each cluster, the delay in sub-trees of each cluster is less than the delay in similar techniques. On the other hand, the proximity of nodes causes the sub-trees not to be under much stress. Therefore, the final tree will have a very low delay and stress.
Keywords
one-to-many content distribution; multicast; peer to peer networks; cluster; stress