Reinforcement Learning Algorithm for Load Balancing in Self-Organizing Networks

Self-Organized Networks (SON) are a crucial feature to make 5G more efficient in 3GPP Release 16. 5G SON inherits from LTE self-configuration, self-optimization, and self-healing functions, and also expands to new use cases and relies on a service-based architecture and data analytics network functions. SON algorithms are, however, not part of the standard. In this article, we thus present a novel reinforcement learning approach for distributed load balancing in heterogeneous networks that use cell range expansion for user association and Almost Blank Subframe (ABS) for interference management. We model the interactions among the base stations for load balancing as a near-potential game. Using the proposed distributed learning algorithms, players reach an optimal Pure Nash Equilibrium (PNE). We provide sufficient conditions under which the learning algorithms converge to the optimal PNE. By running extensive simulations, we show that the proposed algorithms converge within few hundreds of iterations. Finally, we show that outage can be controlled and a better load balancing can be achieved by introducing ABS.