Improving Energy Efficiency and Battery Level in Energy Harvesting IoT URLLC System Using DRL

As the Internet of Things (IoT) continues to expand rapidly, Beyond 5G (B5G) applications face the critical challenge of supporting a massive number of connected devices while meeting Ultra-Reliable and Low-Latency Communication (URLLC) requirements. This article addresses the need for optimizing Energy Efficiency (EE) in such environments by proposing a strategy that minimizes co-channel interference and manages energy harvesting through an adaptive time-switching mechanism. An Iterative Approach Algorithm (IAA) is introduced to dynamically adjust time-switching parameters, effectively reducing interference, enhancing battery performance, and maximizing EE. Furthermore, Deep Reinforcement Learning (DRL) techniques are employed to improve learning efficiency, reduce transmission errors, and enhance both network reliability and Quality-of-Service (QoS). The proposed DRL-based decision selection framework ensures strict compliance with URLLC and QoS demands while optimizing energy usage. Simulation results demonstrate that the energy harvesting system, starting with a 35% battery level using a 700W shell, quickly achieves higher processing capabilities, leading to more efficient power optimization and sustained performance in dense IoT environments.