Content Distribution Strategy Using Web-Cached Multicast Technique

Abstract

In this paper, we propose content distribution strategy to evenly disperse traffic over network and to reduce the required bandwidth for transmitting content data by merging the adjacent multicasts depending upon the number of proxies n that have requested the same one. In our technique, streaming for the identical content is fragmented as long as the grouping interval for batching multicast and can be stored among proxies in order of the requests. A client might have to download data on two channels simultaneously, one from server through multicast and the other from proxies through unicast or multicast. According to the popularity of content, the grouping interval of multicast can be dynamically expanded up to n times and so it can be reduced server’s workload and network traffic. We adopt the cache replacement strategy as LFU (Least-Frequently-Used) for popular content, LRU (Least-Recently-Used) for unpopular content, and the method for replacing the first block of content last to reduce end-to-end latency. We perform simulations to compare its performance with that of conventional multicast. From simulation results, we achieve that the proposed content distribution strategy offers significantly better performance.