Cross-tier interference mitigation in Femto-macro cellular architecture in downlink

Deployment of femtocells faces a significant challenge emanating from the interference from the overlay of macros to the underlay of femtos. This paper focuses on the mitigation of this cross-tier interference in the downlink by limited coordination between femtocells and macrocell. Contesting the natural strategy of orthogonalized transmission thereby eliminating the interference, we propose a spectrum splitting strategy which is based on coordinating this cross-tier interference in a way that the resultant interference is effectively exploited thereby improving the system performance. The problem is posed as a single-agent control problem. Macro base station (BS) is an agent that dynamically swaps its macro users/constellations on its bandwidth to manage the interfering constellations that it produces to its femtocell users. The agent learns the optimal interfering constellation strategy using information from its femto BSs using reinforcement-learning with a Q-learning implementation. Simulation results illustrate the sum rate gain brought about by the proposed strategy of managing interfering constellations for their subsequent exploitation.