A Distributed Protocol for Wireless Sensor Networks Based on Multiple-Leader Stackelberg Network Games

While the past literature on game theory rarely addresses the implementation of protocols that converge to Nash equilibria, practical networks must be designed with protocols that correctly address issues of information asymmetries, hysteresis effects due to these asymmetries, and the fact that information can propagate only locally on networks. To this end, we develop a distributed protocol based on multiple-leader Stackelberg network games to efficiently utilize the localized energy resources of a sensor network. Our protocol arrives at a Nash equilibrium of the multiple-leader Stackelberg network game. We demonstrate the performance of our protocol under a large-scale fading model for the internodal links, and quantify its control overhead. Through this work, we find that there are considerable differences between centralized implementations of algorithms that find Nash equilibria on networks, and distributed protocols that can converge to those equilibria in practice.