Distributive wireless network resource allocation with nash equilibrium and internal-regret-learning of non-stationary actions

This paper presents an iterative method in solving distributive wireless network resource allocation at the shared link with bottleneck. We propose a utility function considering trade-off between transmission bit rate and power efficiency. Given other players´ transmission strategies, the utility function of each player is a concave function. Next, we formulate resource allocation problem as a game, where each player compete to use network resource under its own power constraint. All players utilize the Modified Internal-Regret-Learning algorithm to find their own transmission strategies, which finally form a Nash equilibrium point. The convergence and rate of convergence of the proposed algorithm are proven. Then, we study the results of distributive resource allocation under partial knowledge of other players´ strategies. Simulations are conveyed to show the results of resource allocation under various setup environments.