Enforcing cooperation in ad hoc networks with unreliable channel

An inherent assumption for packet forwarding in ad hoc networks is that the nodes will cooperate i.e., nodes can rely in each other. Thus, it is extremely important that cooperation is induced and achieved in the network. In this paper, we use game theory to analyze the necessary and sufficient conditions to enforce cooperation enforced, especially when a node cannot perfectly monitor other nodespsila behaviors. We analyze a credit exchange method under a general unreliable channel and show that the packet forwarding probability can be adjusted through proper design of incentives, which in turn can be used to attain the desired Nash Equilibrium. We extend our discussion to repeated games and take several well-known strategy profiles and derive the conditions under which the cooperation can lead to a subgame perfect Nash equilibrium. In particular, we show how the unreliable channel can affect the conditions and how a reputation based strategy leads to subgame perfection even under imperfect monitoring. We further investigate collusion resistance and cooperation coalition formation using evolutionary game theory. Mathematical proofs show the existence of an upper bound on the population share of the non-cooperative nodes for an evolutionarily non-stable strategy that enforces full cooperation. This bound is shown to depend on the nodespsila belief on the continuity of the game.