Resource Allocation in Information-Centric Wireless Networking With D2D-Enabled MEC: A Deep Reinforcement Learning Approach

Recently, information-centric wireless networks (ICWNs) have become a promising Internet architecture of the next generation, which allows network nodes to have computing and caching capabilities and adapt to the growing mobile data traffic in 5G high-speed communication networks. However, the design of ICWN is still faced with various challenges with respect to capacity and traffic. Therefore, mobile edge computing (MEC) and device-to-device (D2D) communications can be employed to aid offloading the core networks. This paper investigates the optimal policy for resource allocation in ICWNs by maximizing the spectrum efficiency and system capacity of the overall network. Due to unknown and stochastic properties of the wireless channel environment, this problem was modeled as a Markov decision process. In continuousvalued state and action variables, the policy gradient approach was employed to learn the optimal policy through interactions with the environment. We first recognized the communication mode according to the location of the cached content, considering whether it is D2D mode or cellular mode. Then, we adopt the Gaussian distribution as the parameterization strategy to generate continuous stochastic actions to select power. In addition, we use softmax to output channel selection to maximize system capacity and spectrum efficiency while avoiding interference to cellular users. The numerical experiments show that our learning method performs well in a D2D-enabled MEC system. 2020 Association for Computing Machinery. All rights reserved. This work was supported in part by the National Natural Science Foundation of China under Grant 61772387, in part by the Fundamental Research Funds of Ministry of Education and China Mobile under Grant MCM20170202, in part by the National Natural Science Foundation of Shaanxi Province under Grant 2019ZDLGY03-03, in part by the Graduate Innovation Fund of Xidian University under Grant 5001-20109195456, and in part by the ISN State Key Laboratory. Scopus 2-s2.0-85077531009