Energy Efficient Multi Hop D2D Communication Using Deep Reinforcement Learning in 5G Networks

Abstract

One of the most potential 5G technologies for wireless networks is device-to-device (D2D) communication. It promises peer-to-peer consumers high data speeds, ubiquity, and low latency, energy, and spectrum efficiency. These benefits make it possible for D2D communication to be completely realized in a multi-hop communication scenario. However, the energy efficient multi hop routing is more challenging task. Hence, this research deep reinforcement learning based multi hop routing protocol is introduced. In this, the energy consumption is considered by the proposed double deep Q learning technique for identifying the possible paths. Then, the optimal best path is selected by the proposed Gannet Chimp optimization (GCO) algorithm using multi-objective fitness function. The assessment of the proposed method based on various measures like packet delivery ratio, latency, residual energy, throughput and network lifetime accomplished the values of 99.89, 1.63, 0.98, 64 and 99.69 respectively.