AN IMPROVED TRUST MODEL FOR MITIGATING DISHONEST RECOMMENDATION PROBLEM IN MANET

Abstract

MANET is a wireless network with mobile nodes without having any prior infrastructure setup. MANET does not require any access point, router, server etc. The nodes of the MANET communicate themselves through intermediate hop nodes. Each node in MANET acts both as host and also router. This is a self-organized and self-configured network. They use wireless channels randomly for communication. MANETs have many different features as compared to infrastructure network like dynamic topology, nodes higher mobility, low bandwidth, low battery power, higher error rates etc. The nodes in MANET can be connected anywhere any time arbitrarily due to its dynamic features. Also nodes in MANET act as a router and participate in route discovery and maintenance. Security is a major issue in MANET. MANET is vulnerable to different types of attack due to its features like dynamic topology, limited bandwidth, battery power etc. The attackers can attack in various ways, like sending fake routing information, fake messages, flooding packets, flooding false packets etc. There are various types of attack in each layer of the network. Different trust models have been developed for securing the network using recommendations from other nodes, but dishonest recommendations may lead to take wrong decision. This paper presents a probability based trust model which uses recommendations from other nodes. Here, the possible attacks that can occur due to dishonest nodes while taking recommendation are investigated. Here, a filtering algorithm is proposed for resisting the effect of misbehaving nodes while taking recommendations for selecting valid nodes for recommendation.