Adaptive biasing scheme for load balancing in backhaul constrained small cell networks

In this study, a distributed biasing scheme is designed to achieve load balancing for heterogeneous networks. Based on the limited backhaul capacity and user distribution in the system, each small cell base station adaptively and distributively changes its cell range by setting the bias value, to effectively utilise the wireless resource and achieve load balancing as well. The Q-learning algorithm is adopted to design the biasing scheme in each small cell base station. The tradeoff between the backhaul resource utilisation and the quality-of-service of users is considered in the reward function of the Q-learning model. To examine the performance of the distributed scheme, a centralised scheme aiming at maximising the backhaul resource utilisation is also proposed for comparison, whose performance lower bound is derived. Numerical results show that the proposed distributed scheme can effectively utilise the backhaul resource for load balancing, and achieve a close performance to the centralised one.