Mahdi Sattarivand - School of Electrical and Computer Engineering, Guilan University, Guilan, Iran

Abstract— Peer-to-Peer ( P2P ) 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.Index Terms — peer - to - peer systems , confidence , genetic programing , malicious nodes .

peer-to-peer systems, confidence, genetic programing, malicious nodes