HPM: A novel hierarchical Peer-to-Peer model for lookup acceleration with provision of physical proximity

Abstract

It is well known that Peer-to-Peer systems are generally featured with high flexibility and scalability, enabling dynamic resources localization and mutualization, and allowing the nodes to freely join and leave. But in some special environments such as mobile P2P networks, routing optimization, resources reliability and availability are critical concerns. To deal with these issues, we propose a novel architecture, termed hierarchical Peer-to-Peer model or HPM for short, based on Chord for improving P2P network performance in the presence of such additional requirements as fault tolerance and self organization. Specifically, HPM is composed of a set of hierarchical rings, each of which consists of the nodes that are both physically and logically close to each other or we say they have physical proximity, supporting inter and intra routing mechanisms. We show that the cost of lookup for HPM is O(∑i=14log2(ni)), where ni represents the number of nodes on ring level i (with the maximum of 256 nodes in each ring in the case of four levels). Each node maintains a routing table with only 2×O(log2(ni)) entries, greatly facilitating HPM to work in resource-limited terminals such as mobile phones or PDA. In particular, when HPM is combined with a broadcast mechanism, the lookup process can be significantly improved (four hops). Our simulations and comparative studies demonstrate that HPM can achieve satisfactory P2P performance with rapid convergence thanks to the cost-optimal lookup algorithm.