Building Efficient Limited Scale-Free Overlay Topologies for Unstructured P2P Networks

In addition to the peer/data organization and search algorithm, the overlay network topology (logical connectivity graph) among peers is a crucial component in unstructured peer-to-peer (P2P) networks. Its topological characteristics significantly affect the efficiency of a search on such unstructured P2P networks. Scale-free (power-law) overlay network topologies are one type of the structures that offer high performance for these networks. However, a key problem for these topologies is the high connectivity (i.e., load) in only a small portion (i.e., hubs) of nodes. In fact, the peers in a typical unstructured P2P network may not want or be able to cope with such high connectivity/load. Therefore, some hard cutoffs are often imposed on the number of edges that each peer can have to achieve implementation feasibility and fairness among all peers, resulting in limited or truncated scale-free networks. In this paper, we discuss and analyze the growth of such limited scale-free networks and propose two different algorithms for constructing perfect scale-free overlay network topologies with flexible scale-free growth exponent (γ) and with low communication overhead during the construction phase. Through extensive simulations, we also evaluate the performance of proposed approaches and compare them with existing solutions. The simulation results demonstrate that the proposed approaches indeed generate perfect scale free networks and can achieve the best performance in different search algorithms with right parameters with much smaller overhead than the existing solutions.