Federated Learning Enabled Green Edge Computing System for IIoT Applications

These days, the usage of industrial Internet of Things (IIoT), such as industrial robotics, smart homes, and electric, has been growing progressively. The IIoT applications consisted of different sensors and generated huge amounts of data. Due to resource issues of local devices, generally, IIoT workloads offload to a centralized cloud for execution. However, centralized power consumption is much higher, which is not optimal in a green environment. Green edge nodes are an extended version of cloud computing that brings computing devices and infrastructure to the network’s edge. They are designed to be environmentally friendly and energy-efficient. However, edge node-enabled green systems still have many issues. This paper presents a novel conceptual federated Learning Enabled Green Edge Computing system for IIoT Applications. The paper aims to reduce the power consumption of computing nodes such as local devices, edge nodes, and cloud nodes during data training and analysis of IIoT applications. The proposed system offers green data training and analysis services at the edge nodes and offloads the data to the cloud for storage. The paper presents energy-green efficient federated learning deep convolutional neural network scheduling (FL-DCNNS) algorithm methodology to schedule all workloads of IIoT applications with minimum power consumption. Simulation results show that FL-DCNNS meets the requirements of a green computing environment on edge nodes during the execution of IIoT applications as compared to existing methods.