Security mechanisms in mobile adhoc networks

Abstract

Mobile Adhoc networks (MANETs) are accepted effortlessly worldwide due to their potential of bringing low cost mobile connectivity solutions to everyday communication problems [1-2]. These networks have been widely applied in many commercial and military applications. Therefore, security of these network has become imperative and vital challenge. The main goal of the thesis is to design a robust, efficient and accurate method for providing security in MANETs. In this regard, various security methods have been studied and efforts has been made to find out appropriate solution which can provide security in a given mobile adhoc networks. Although various security methods are available in literature, but all have their own advantages and disadvantages. Thus, the objective is to find more suitable method for MANETs application and evaluate it in terms of network throughput, accuracy, delay, energy and time required for detection of an attack in any given network. The watchdog is a well-known Intrusion Detection System (IDS) for MANETs, which not only monitors the traffic between peer nodes but also performs analysis on the data to discern any malicious activity in the given network. It has been adopted for detecting network layers attack i.e. black-hole attacks, worm-hole attack, flooding attack etc. But Watchdog method suffers from serious limitation viz. high level of false positive and negative of attack detection. Integration of the Bayesian filtering in watchdog, improves attack detection. And, the Bayesian-watchdog intrusion detection system can be further enhanced by its effective tuning of the Bayesian filter. This research presents the concept of trust-based-tuning of the Bayesian-watchdog, which is a novel approach towards enhancing the detection speed, eliminating false alarms and improving data throughput. The proposed trust based tuning algorithm dynamically tunes the Bayesian filtering function according to the environment and network conditions. The proposed algorithm has been evaluated ...