Cluster Based Wireless Sensor Network Security Model Using Game Theory and Risk Assessment

Abstract

The purpose of this thesis is to model an appropriate response to an attack in a wireless sensor network by implementing game theory concepts coupled with risk assessment. In contrast to the widely analyzed key management techniques which involve computation and communication overhead, the security model is designed as a non-cooperative game between the attacker and the network. By creating and using a model that uses a risk assessment model to quantify risk and employing a cost-payoff relation of the game theory for decision making better resource utilization and better security can be achieved. The risk assessment model integrated with a game theoretic framework is investigated for efficiently securing a wireless sensor network. A new methodology to effectively estimate risk based on the threat and damage values, using fuzzy logic has been introduced. The cost of defending a node and the related gain in payoff with respect to the objective of the network was considered while analyzing the game. The game model reduces the computational overhead with nodes by moving the decision making to cluster heads. By choosing appropriate utility parameters and payoff function the network can secure itself efficiently.