A Semi-Distributed Scheme for Mode Selection and Resource Allocation in Device-to-Device-Enabled Cellular Networks Using Matching Game and Reinforcement Learning

Device-to-Device (D2D) communication is a promising technological innovation that is significantly considered to have a substantial impact on the next generation of wireless communication systems. Modern wireless networks of the fifth generation (5G) and beyond (B5G) handle an increasing number of connected devices that require greater data rates while utilizing relatively low power consumption. In this study, we present joint mode selection, channel assignment, and power allocation issues in a semi-distributed D2D scheme (SD-scheme) that underlays cellular networks. The objective of this study is to enhance the data rate, Spectrum Efficiency (SE), and Energy Efficiency (EE) of the network while maintaining the performance of cellular users (CUs) by creating a threshold of data rate for each CU in the network. Practically, we propose a centralized approach to address the mode selection and channel assignment problems, employing greedy and matching algorithms, respectively. Moreover, we employed a State-Action-Reward-State-Action (SARSA)-based reinforcement learning (RL) algorithm for a distributed power allocation scheme. Furthermore, we suggest that the sub-channel of the CU is shared among several D2D pairs, and the optimum power is determined for each D2D pair sharing the same sub-channel, taking into consideration all types of interferences in the network. The simulation findings illustrate the enhancement in the performance of the proposed scheme in comparison to the benchmark schemes in terms of data rate, SE, and EE.