A game-theoretical model for intrusion detection in wireless sensor networks

Abstract

In this paper, we propose a game-theoretical based method to fortify an intrusion detection system (IDS) in Wireless Sensor Networks (WSNs). And we focus on dropping-packet attacks which are the sore point in wireless sensor networks. We claim that of course our model is not only a detective model but also a correctional model. because first we corner malicious nodes and then we give them incentives to mend their ways and try to behave in favor of other nodes in WSN. We formulate the attack-defense problem as a non-cooperative, two-player, non-zero-sum game between an attacker and a WSN. We propose a repeated game model of dropping packets which is attack-proof against DPA based on such an assumption that sensor nodes are rational. The model prevents malicious nodes from attacking by establishing subgame perfect periodic collusion-resistant punishment mechanism, and impels sensor networks to reach a cooperative Nash equilibrium. So, it leads to a defense strategy for the network. And when the malicious nodes populate less than the majority of network we show that our model is partially collusion resistant. And even we try to use the sporadic well-mannered cooperative behavior of the malicious node in favor of improving overall throughput of our WSN. Based on rationality we believe that even malicious nodes before being known as extreme notorious nodes, they try to impersonate their true type and show a little collaborative behavior.