Two-phase exclusion based broadcast adaptation in wireless networks

Abstract

The basic idea of popularity-based wireless broadcast is to broadcast data with most requests. However, client requests cannot always reflect their entire data requirements. It thereby leads to the inaccuracy of broadcast and the increased number of client requests. In this paper, we propose a correlationbased broadcast scheme, in which relationships among data are used as a reference in dealing with exclusion of unpopular data. In this broadcast scheme, data to be excluded are initially divided into two parts. The popularity of data in the second part is inferred by the actual popularity of data in the first part. This is possible if a causal relationship exists in their access patterns. The results from extensive simulation experiments conclude that the correlation-based broadcast scheme can significantly improve the mean response time and reduce the number of missing requests. © Springer-Verlag Berlin Heidelberg 2005.