Optimized Performance Evaluation of a Q-Learning Hard Handover Algorithm for Load Balancing

This paper presents a method for load balancing of cellular base stations based on self-organized networks and the Q-learning algorithm (reinforcement learning). The aim of the method is to redirect the load from an overloaded service cell to a non-overloaded neighboring cells. Simulation results show that the higher the value of the handover margin variation step, the greater the difference in the handover margin variation so the maximum reward was obtained between the different learning rate values.