TAP: Topology-aware peer-to-peer network with expanding-area lookup

Abstract

To solve the scalability problem of unstructured Peer-to-Peer(P2P) network, Distributed Hash Table(DHT)-based P2P network is proposed. DHT-based mechanism, however, does not reflect physical topology of P2P network, so that messages are routed inefficiently. As a result, it makes long latency of lookup and overload in networks. In this paper, we propose a new Topology-Aware Peer-to-Peer(TAP) mechanism which adopts DHT-based approach but solves the problem of inefficient routing of lookup messages. The key idea of TAP is to generate node identifier (NodeId) and object identifier (ObjectId) with hierarchical topology information like country, province and organization. An object is inserted to each level of area and then, nodes lookup the object with expanding area gradually. It leads to fast lookup and download which is 10 similar to 20 times and 2 times faster than Chord respectively in our experiment. In addition, we can acquire some features like locality, cache-like effect and local broadcast.