Fog Intelligence for Energy Optimized Computation in Industry 4.0

The field of communication and computation is experiencing ongoing expansion with the emergence of Industry 4.0 technology enabling efficient data transfer among devices. However, this advancement also poses several challenges, particularly in managing the vast amount of data generated by Industrial Internet of Things (IIoT) devices. Despite being widely recognized as an effective solution to these challenges, cloud computing also poses its own challenges. These include high bandwidth usage, latency, security concerns, and energy dissipation. In an effort to mitigate these issues, fog computing has emerged as a more energy-efficient alternative. The primary focus of this paper is the reduction of energy consumption in industrial fog networks. To accomplish this, we propose a novel architecture with the integration of fog networks and Deep Reinforcement Learning (DRL) technique to optimize the overall system reward and reduce energy consumption in industrial applications. The problem of state-action-reward is formulated as a Markov Decision Process (MDP) and optimized using a popular DRL technique. Simulated results indicate that the proposed strategy decreases energy consumption rate by 10% compared to existing offloading strategies by offloading decisions on different computing devices.