Designing Time Difference Learning for Interference Management in Heterogeneous Networks

In this paper, we model the interference management problem in heterogeneous femto and macro networks, through a stochastic game. We claim that in a realistic wireless scenario this game cannot be analytically solved, so that we propose a solution based on a Reinforcement Learning (RL,) scheme. We present a taxonomy of RL, approaches, and we propose to select the most appropriate one to find a solution to our problem. Once we select the most adequate learning method, we study how to optimally design it in order to maximize the system performances.