On optimal and cheat-proof packets forwarding strategies in autonomous ad hoc networks

This paper mathematically studies the cooperation and packet forwarding issues among selfish nodes in ad hoe networks under a game theoretic framework. Since in such packet forwarding games there usually exist an infinite number of Nash equilibria, a critical issue is how to perform equilibrium refinement, that is, how to apply extra optimality criteria to remove those equilibrium strategies that are less robust, less rational, or less likely. In this work, besides Pareto optimality and subgame perfection, other important optimality criteria, such as social welfare maximization, absolute fairness, and proportional fairness, have also been considered when performing equilibrium refinement. Combining with Pareto optimality and subgame perfection, each of these criteria can lead to a unique Nash equilibrium solution. Since nodes are selfish and will try to cheat whenever possible, the possible cheating behavior has also been fully exploited, and the analysis has shown that when cheating behaviors are considered, all these unique equilibrium solutions will converge to the same format, that is, a node should not help its opponent more than its opponent has helped it.