Modelling Bayesian attacker detection game in wireless networks with epistemic logic

In collaborative networks operating over a wireless medium, to maximize connectivity of the network, two nodes that are not directly connected may need to communicate with one another through other nodes in the network. However in such networks, not all nodes are reliable. Therefore it may be necessary for legitimate users to monitor the behavior of other nodes such that they can form a belief about whether those nodes are potential attackers or not. These beliefs can be studied precisely in game theoretic scenarios such as Bayesian attacker detection games. These games are strategic games with incomplete information for modeling the interaction between nodes in wireless networks with channel uncertainty. However these games do not normally consider the reasoning abilities of agents. One solution is to use formal logics which allow for modeling the reasoning abilities of agents as well as allowing formal reasoning about certain properties of games such as the Nash equilibrium. In this paper, we propose an epistemic logic to study strategic games with incomplete information. In this logic we can precisely describe what the requirements and consequences of informative actions are. We can reason what rational agents should do if they can choose between different available actions. In addition, this language can be used as a semantically well-defined query language for model checkers to automatically verify the game descriptions with respect to their intended specifications.