A Fast Algorithm for Construction of Minimum Delay Multicast Trees in P2P networks

Abstract

Peer to Peer networks have received considerable attention during the past decade and have been the focus of many research and development activities. New applications such as video streaming or gaming would benefit from multicast topologies and creating such trees in peer to peer networks is a challenging task. Proper multicast trees should minimize the traffic delay and bandwidth consumption. At the same time, construction of these trees should not be very complicated in terms of the required processing power so they can be implemented easily. The required time for construction of multicast trees should also be minimized and that helps with their adaptation to the dynamics of the network. Proposed approaches in the literature are either very complicated to implement by ordinary nodes in P2P networks, or unable to reach the optimal tree noting the delay and bandwidth parameters. In this paper we propose a novel algorithm that can be used to construct the minimum delay multicast tree considering nodal bandwidth constraints. The algorithm is fast, simple and provides considerable improvement over the current proposed techniques. To achieve these advantages, we used a new tree representation by which the network is modeled. The optimal tree is then sought by ordinary and co-evolutionary genetic algorithms. Simulation results on networks with 16 to 200 nodes are reported and compared for our proposed method and QoSMR_GA.