Survey on Reinforcement Learning Applications in Communication Networks

Abstract

In recent years, intelligent communication has drawn huge research efforts in both academia and industry. With the advent of 5G technology, intelligent wireless terminals and intelligent communication networks are increasingly under intensive study. Artificial intelligence enhances the network capability with automatic and adaptive adjustment. Reinforcement learning (RL) and deep reinforcement learning (DRL) are two powerful techniques in artificial intelligence which can learn the optimal decision according to the environment feedback. In this paper, we focus on the latest research progress on RL and DRL applied in three emerging technologies including mobile edge computing (MEC), software defined network (SDN) and network virtualization in 5G. The prospect of further research and development in the future is preliminarily forecasted.