Game theoretic approach to model node misbehaviour in MANETS

Abstract

The security problems in Mobile Adhoc Networks (MANETs) are being constant topic of interest and drawn attention of research community from past decade, where massive study has already been presented for addressing the same. One of the precarious problems encountered in MANET is that of identifying the malicious nodes. The identification and later mitigation of the same becomes immensely difficult task especially when Selfish / Erroneous nodes exist along with normal Collaborative nodes in the Regular camp. The presence of selfish nodes is potentially harmful as similar behaviour can be imitated by malicious nodes which are the point of concern of many security aspects. While performing the preliminary study of the proposed system, it was known that large amount of previous study has emphasized on usage of cryptographic protocols to secure MANET, where implementation becomes a big question mark and challenges time and space complexity of those sophisticated algorithm. While reviewing some more techniques on security system in MANET, it was found that game theory has also prime contribution in past few years due to potential accuracy in its probabilistic approach and computational efficiency. Therefore, the proposed system accentuates the use of game theory and probability theory considering selfish nodes in the regular node camp while modeling the Regular versus Malicious node game and thereby enhancing the prior mathematical schema of strategical decision making to accommodate for the same. The study performs statistical analysis and presents a mathematical model to mimic the multi-stage game between regular and malicious node using Game Theory. The framework effectively represent the various unpredictable actions of node cooperation, node declination, node attacks , as well as node reporting that can model the strategic profiling of various mobile nodes. Understanding the patterns and then deploying the algorithms in security products can reduce intrusion to a greater extend. The simulation has been conducted in Matlab for monitoring the trends of pure strategy, mixed strategy, and PBE strategy adoption to visualize the behavioral pattern of mobile nodes. The outcome of the study shows significant enhancement in reporting actions of regular node thus considerable reduce down vulnerability scenario in MANET. The study lays emphasis on Perfect Bayesian Equilibrium (PBE) strategy which forms as the basis of all the result analysis. The enhancement was found in terms of 62.67 % lesser false positives which favors higher overall network utility (modeled as utility of regular nodes in the game) with Selfish / Erroneous nodes existing in the network when collating the proposed schema with prior work.