Improving the routing performance of KAD through social network analysis

Peer-to-peer (P2P) file-sharing networks are gradually becoming popular. Among them, eMule´s KAD, a network deployed to publish and search files for eMule, is a typical example. Despite the popularity of KAD, its routing performance still has much room for improvement. This paper presents a novel approach to improve the routing performance of KAD through social network analysis. We propose a novel metric, DOC, to measure the degree of concentration of IP addresses of peers in a community (a group of peers which share some files). Based on DOC, we find that IP addresses of peers in a community cluster together. Based on this characteristic, a better routing table maintenance algorithm, RTMA-2, is proposed to improve the routing performance of KAD. Our theoretical analysis shows that RTMA-2 yields a significant performance improvement over the original routing table maintenance algorithm, RTMA-1.