SPTrust: Reputation Aggregation Method Based on Similarity to Reputation Scores of Power Nodes in Unstructured P2P Networks

Abstract

Reputation aggregation methods are used in unstructured peer-to-peer (P2P) networks to evaluate the trustworthiness of participating peers and to combat malicious peer's behaviors. In reputation aggregation methods, each peer collects local scores by each transaction and calculates global scores by aggregating local scores. In each transaction, global scores enable peers to interact with reliable peers. Gossip Trust is proposed as a reputation aggregation method for the unstructured P2P networks. This method refers to reputation scores of power nodes, and power nodes are the peers of the high-ranking global scores. Although, there are dishonest peers that forge reputation scores of their own against other peers but get high global scores by providing authentic files in the networks. Gossip Trust does not consider the influence of forged reputation score when dishonest peers exist and are selected as power nodes. In this paper, we propose a reputation aggregation method called SPTrust. SPTrust is based on the similarity to reputation scores of power nodes. In SPTrust, each peer calculates the similarity value to reputation scores of power nodes. And it can detect that dishonest peers are selected as power nodes. By using SPTrust, we can effectively decrease the influence of forged reputation score from malicious peers and solve the problem in Gossip Trust. In computer simulations, SPTrust has been shown to decrease the number of inauthentic files downloads compared with Gossip Trust.