An evolutionary game approach to P2P video streaming

Abstract

This paper deals with the allocation of resources and the creation of distribution trees for P2P video streaming. Assuming that the video distribution is based on a Multiple Description Codec (MDC), the proposed approach performs jointly the two tasks, in a completely distributed client-based fashion, by means of an evolutionary (socially inspired) game played by all peers. The key idea of the evolutionary game is that each peer continuously measures its own utility (i.e., video quality) and, by periodical comparison with other randomly chosen peers, tries to mimic the behavior (i.e., resource allocation and video distributors) of peers with higher utility. Extensive simulation experiments by Peersim have been carried out in order to assess the performance of the proposed technique. The obtained results have highlighted remarkable benefits in terms of scalability, fast adaptation to churning, high degree of cooperation among peers and self-organization of the distribution multitree network.