A Genetic Programming-based trust model for P2P Network

Peer-to-Peer systems have been the center of attention in recent years due to their advantage. Since each node in such networks can act both as a service provider and as a client, they are subject to different attacks. Therefore it is vital to manage confidence for these vulnerable environments in order to eliminate unsafe peers. This paper investigates the use of genetic programing for achieving trust of a peer without central monitoring. A model of confidence management is proposed here in which every peer ranks other peers according to calculated local confidence based on recommendations and previous interactions. The results show that this model identifies malicious nodes without the use of a central supervisor or overall confidence value and thus the system functions.