Deep Reinforcement Learning for Intelligent Communications

In the era of data explosion, the rapid growth of mobile devices makes the size of wireless networks increase tremendously. Meanwhile, people are having higher demands for wireless communications, which requires the networks to provide precisely on-demand services in order to exploit the limited resource. Due to the above two reasons, the traditional modeling-and-optimizing methods for network management will meet the performance bottleneck in the future. Fortunately, the appearance of artificial intelligence and machine learning provides a new solution to this issue. As a data-driven machine learning technique, deep reinforcement learning can directly learn the pattern of dynamic environments and use it to make optimal decisions. Hence, deep reinforcement learning enables wireless networks to manage and optimize themselves based on their environments, which makes it possible to realize intelligent communications. This paper introduces the application of deep reinforcement learning on wireless communications from the aspects of resource management, access control, and network maintenance, and illustrates that deep reinforcement learning is an effective approach to realizing intelligent communications. © 2020, Editorial Board of Journal of the University of Electronic Science and Technology of China. All right reserved.