P2P multicasting network design problem — Heuristic approach

Abstract

Streaming services have been gaining much popularity in the Internet. The widespread adoption of broadband residential access enables individual users to use new bandwidth demanding services. Peer-to-Peer (P2P) multicasting approach using an overlay network to provide communication between peers is a valuable proposal for streaming since it provides rapid deployment at relatively low cost. Most of previous research on optimization problems related to P2P multicasting is limited only to optimization of network flows. In this work we focus on a new problem of P2P multicasting network design, i.e. flows and links' capacity are optimized jointly. We assume that the P2P multicasting system is static with low membership change rate (e.g. corporate videoconferencing, distance learning, delivery of important messages). The objective is to minimize the overlay network cost. We propose an effective heuristic algorithm. Numerical experiments are run in order to examine the proposed method against optimal results and Lagrangean relaxation algorithm as well as to illustrate the dependency of the network cost on various parameters related to P2P multicasting. The proposed algorithm can be applied to a wide range of overlay network design problems including not only streaming services but also computing and storage systems using the overlay network approach.