Intelligent user association algorithm in heterogeneous wireless cellular networks

Abstract

In heterogeneous cellular networks, user association is the key for network performance. In addition, reasonable bandwidth allocation of backhaul network can make full use of network resources and thus improve network performance. In view of this, the authors propose an intelligent user association algorithm considering backhaul bandwidth allocation to maximise the weighted sum rate of network while guaranteeing the requirements of quality of service and avoiding the congestion of backhaul network. Firstly, they predict traffic of each base station using adaptive wavelet decomposition and back-propagation neutral network, which considers the tradeoff between accuracy and complexity. The appropriate wavelet decomposition level can be adaptively selected according to the different mean and variance of traffic. Then, backhaul bandwidth is allocated to different base stations according to the predicted results. Finally, they adopt a method based on penalty function and Marks-Wright algorithm, which transforms the mixed integer programming problem into a series of convex optimisation problems that can be solved iteratively, to complete the user association. The simulation results show that the prediction algorithm obviously reduces error and the association algorithm has fast convergence speed.