Optimal cache allocation algorithm using threshold-based multicast

Abstract

VoD is generally transmitted by unicast to meet its on-demand characteristic, however, when the requests are concentrated on a video, unicast works inefficiently. To take advantage of multicast, patching has been designed. Patching transmit a video using a multicast channels and patch missed data (passed portion before user request) using additional unicast channel. One of patching scheme, threshold-based multicasting opens new multicast stream at an interval time T to maximize the benefit of patching. To enhance the transmission efficiency, various patching schemes and related applications have been developed and need to be developed. Currently, the proxy cache storages placed nearby the end-users are used with threshold-based multicast for cost reduction. By transmit multicast stream at main server and transmit unicast stream at proxy server, streaming load is divided and shared. However, there was no consideration about the tradeoff between the network channel cost and the proxy storage cost while the threshold-based multicast is working with proxy cache storage. As a result, the number of network channel is not reduced effectively especially for the popular contents, and the cost reduction still not optimized. In this thesis, we modify the threshold-based cache allocation algorithm by considering the tradeoff relationship between the transmitted multicast channel cost and the cache cost on proxy server for cache. After that, we find the optimized proxy cache storage amount which results minimum cost. The proposed algorithm utilizes fixed length caching for popular videos, and the existing caching mechanism using threshold-based multicast for the remaining normal videos. The proposed algorithm achieves reduction of total system cost compare to the simple adaption of unicast patching stream of threshold-based multicast to a proxy server