A game-theoretic framework with reinforcement learning for multinode cooperation in wireless networks

In this paper, a game-theoretic framework based on the iterated prisoner´s dilemma (IPD) is proposed to model the repeated dynamic interactions of multiple source nodes when communicating with multiple destinations in ad-hoc wireless networks. In such networks where nodes are autonomous, selfish, and not familiar with other nodes´ strategies, fully cooperative behaviors cannot be assumed. Thus, a Q-learning algorithm is proposed to allow network nodes to adapt to and play the IPD game against opponents with a variety of known and unknown strategies. Simulation results illustrate that the proposed Q-learning algorithm allows network nodes to play optimally and achieve their maximum expected return values.