Deep Q-Network for User Association in Heterogeneous Cellular Networks

Heterogeneous networks (HetNets) balance the traffic load and reduce the cost of cell deployment, which is considered as a promising technology in next generation cellular networks. Due to non-convexity characteristics, it is very difficult to obtain the optimal strategy for user association problem. This paper proposes a new framework to ensure the long-term overall network utility under the premise of guaranteeing the quality of service of downlink user equipment in downlink HetNets. At the same time, a distributed optimization algorithm based on multi-user reinforcement learning is proposed. In order to solve the problem of large computational load of big action space, the optimal strategy is obtained by introducing the method of deep Q-network (DQN). Simulation results show that DQN has better performance than Q-learning method.