Reliability Optimization in Narrowband Device-to-Device Communication for 5G and Beyond-5G Networks

Abstract

The 5G and beyond-5G (B5G) is expected to be a key enabler for Internet-of-Everything (IoE). The narrowband Internet of Things (NB-IoT) is a low-power wide-area enabling technology introduced by the 3rd Generation Partnership in 5G. The objective of the NB-IoT is to enhance the mobile coverage area by increasing the number of repetitions of control and data packets between user equipment (UE) and the base station/evolved NodeB (BS/eNB). While these repetitions improve data delivery for delay-sensitive applications, they degrade the efficiency of the already resource-constrained IoT system by increasing the system overhead and energy consumption. Moreover, NB-IoT devices in the edge region of the cellular coverage area require more repetitions, which augment energy consumption. In this study, we investigate device-to-device (D2D) communication for NB-IoT delay-sensitive applications, such as healthcare-IoT services, to use two-hop communication instead of using a direct uplink. An optimization problem is formulated to achieve an optimal end-to-end delivery ratio (EDR). In addition, this study incorporates Q-Learning-based reinforcement learning (RL) for the selection of an optimal cellular relay, which assists NB-IoT UE in uploading sensitive data to BS/eNB. The proposed RL-intelligent-D2D (RL-ID2D) communication methodology selects the optimum relay with a maximum EDR, which ultimately augments energy efficiency.

Keywords

• Device-to-device (D2D) communication
• machine learning (ML)
• narrowband Internet-of-Things (NB-IoT)
• reinforcement learning (RL)