A Deep Reinforcement Learning based Mechanism for Cell Outage Compensation in 5G UDN

Ultra Dense Networks (UDN) have become one of the key technologies for 5G wireless communications, which can meet the requirements of high-traffic, high-density wireless terminal access. Compared with LTE, there are a large number of heterogeneous cells in UDN. If a failure occurs and can't be alleviated its effect in time, it will lead to a significant drop in network performance. Therefore, the cell outage compensation (COC) problem in the UDN is very important. Although deep reinforcement learning (DRL) has been applied to many scenarios related to the self-organizing network (SON), there are fewer applications for cell outage compensation. In this paper, aiming at the cell outage scenario in the UDN with the goal of maximizing the sum of the throughput of all users while meeting service quality demands of each mobile user, and present a framework based on DRL to solve it. Specifically, we first allocate compensation users to adjacent BSs by using the K-means clustering algorithm, then apply a deep neural network (DNN) to approximate action-value function. The simulation results show that the algorithm converges quickly and tends to be stable, and reach 99.53% of the maximum throughput. It verifies the efficiency of the DRL-based framework and its effectiveness in meeting the requirement of user rates and handling cell interrupt compensation.