Sustainability and Latency Reduction Through Federated Learning-Powered Digital Twins in Iot Devices

Abstract

The rapid advancement of emerging technologies and the Internet of Things (IoT), including the evolution of Digital Twins (DT), necessitates an accelerated pace in the Beyond Fifth Generation (B5G). This is crucial to establish widespread wireless access by ensuring resilient and immediate wireless connectivity within the real network environment. This article uses edge networks and DTs with blockchain technology. Ensuring robust real-time data processing while providing a scalable and secure solution is the aim. Bridging the gap between digital systems and physical edge networks is the goal. In this research, we bridge the gap between physical edge networks and digital systems by introducing Networks with Digital Twin Edges (NDITE), which combine digital twins and edge networks. Next, we propose a blockchain-driven federated learning method in NDITE to improve data privacy and communication security. We schedule relaying users and manage bandwidth resources using DT-powered Deep Reinforcement Learning (DRL) to increase efficiency. According to the simulation results, the suggested DRL agent-based DT can minimize the weighted cost of transmission policy of edge computing strategies and choose 47.5% of computing tasks to be completed locally with 1 MHz of bandwidth. It can also exploit the optimal policy.