Power Control Algorithm For Femtocell Based On Machine Learning

With the development of the next generation wireless communications technology, theconcept of femtocell appears. Femtocell is a kind of home base station which can expandcoverage indoor, improve system performance and decrease macrocell communicationoverhead. However, the existing co-channel interference of the network between macrocellsand femtocells can dramatically degrade the overall performance of the network. In a two-tiercellular network comprised of a central macrocell underlaid with femtocell access devices, theinterference between layers limits the capacity of the network.This paper analyzes the femtocell base station transmit power control in the downlink andgives the distributed power control algorithm based on Q-learning, and compares independentQ-learning and cooperative Q-learning of system capacity and system fairness. After that itimproves the reward function and operation selection method. This algorithm increases thecapacity of macrocell system, and improves the system fairness and capacity of the totalfemtocell system.In this paper, Matlab simulations a scenario which macrocell system is coexisted withmany femtocells systems, and analysis the interference between each base station and the users,and studies the principle of transmission power control on the femtocell base station. Thenvalidate the method of improving the reward function and operation selection principles, withwhich the system fairness and the capacity of the system are improved efficiently.