Peer Discovery in Peer-to-Peer Anonymity Networks

Many peer-to-peer anonymity protocols have been proposed in recent years. One of the key challenges in designing such protocols is peer discovery, especially in decentralized unstructured peer-to-peer systems. In this paper, we propose a Gossip algorithm for Peer Discovery based upon node Local in-degree (GPDL), using only limited knowledge of the network topology. Our goal is try to discover peers uniformly in the whole network as well as try to discover robust nodes based upon local topology information. The idea behind our gossip is that the larger the degree of a node, the better its robustness. Our simulations show that the network under GPDL algorithm is high clustering and its average diameter is almost not influenced by the node dynamics. With time passing, a node can discover robust nodes and its neighbors are nearly selected uniformly from the whole network. The message overhead produced by GPDL is small.