Self-Organized Routing for Unstructured Peer-to-Peer Networks

Abstract

Overlay topologies are often used in unstructured peer-to-peer networks in order to improve the global performance. Various research works dealing with communication issues of dynamic networks gave much interest to overlay topologies to get a better network organization. In fact, the use of such structures aim to decrease the impact of network dynamics while improving its scalability. Moreover, overlays offer a better management of data flows over the network. Recently, some overlays offering scale-free and/or small-world properties have been proposed. However, most of these works focus on structural aspects and do not give much importance the data carried by peers. In this paper, we show the complementarity of self-organization in overlay topologies and self-organizing search algorithms. We improve an existing approach of scale-free overlay topology by introducing mechanisms to take into account the utility of the information on peers and by a novel ant inspired approach based on foraging. We show how the combination of structural and semantic topological aspects with an ant inspired search algorithm is well adapted especially for rare information in the peer-to-peer network.