Evolutionary game theoretic approach for optimal resource allocation in multi-agent systems

For task completion in distributed environments, a set of resources is required and a group of agents must cooperate in deciding the share each should provide to maximize the system performance. We address the problem from an evolutionary game-theoretic perspective and present a fully distributed algorithm based on local replicator dynamics. By using the optimality condition, we prove the convergence and optimality of our algorithm under connected communication topologies. An illustration example demonstrates the effectiveness of the proposed approach in a dynamical environment.