PreferTrust:an Ordered Preferences-based Trust Model In Peer-to-peer Networks

P2P networks have received more and more attention since published. Now P2P networks have been applied to many kinds of web applications. P2P networks give up the C/S mode in traditional networks and adopt a peer-to-peer principle, as which, all entities in the network are equal, they both provide the services and enjoy the services. However, the openness of P2P networks brings potential safety hazard, which puts the development of P2P networks off.Nowadays, the file sharing system, VOD system, IM system and e-commerce system based on P2P networks become more and more popular. However, some problems still exist, as there are some malicious participants who transmit virus and advertisements in the network. To solve this problem, some scholars bring the concept of trust in social relations to P2P networks. Simulating the trust relations among human beings, the trust relations among peers in P2P networks are founded. Existing researches show that trust model can hold back malicious peers’behavior and solve the trust problem in P2P networks in some degree. However, in most existing trust models, a continuous or discrete single value is applied to represent trust mostly. The trust decision method based on that single value may be too simple.Aiming at solving the two problems in existing trust models, rough trust description and oversimplified trust decision, an ordered preferences-based trust model, preferTrust, is proposed for peer-to-peer networks. First, we define a service as an n-dimensional attribute vector in order to reflect the service provider’s abilities in different attributes in detail. Second, considering the fact that different peers may have different preferences for each attribute of a service, which are usually represented in the qualitative expressions, such as better, worse and so on, we transform those qualitative peer’s preferences to the quantitative ordered preferences with the method of analytical hierarchy process, and based on which to calculate the requester’s expectation vector. Finally, we combine the requester’s expectation vector with each responder’s trust vector to calculate an estimate value for each responder which will help the requester make the final trust decision. The simulation results show that preferTrust is able to improve the trust decision satisfaction and resist malicious attacks effectively and efficiently.