CyTFS: Cyber-Twin Fog System for Delay Efficient Task Offloading in 6G Mobile Networks

Sixth-generation (6G) mobile computing is a wireless, cutting-edge technology that is made possible by the digital interconnectedness of everything (IoE). 6G connection depends on mobile edge and fog computing integration. Due to the mobility of various end users, task offloading in these computing devices is hard and unpredictable. The unexpected mobility of users and dynamic switching of networks from 5G to 6G make Mobile Fog Computing (MFC) benchmarks poor for task offloading. To overcome these inefficient task offloading in dynamic unpredictable 6G mobile environments, this paper presents an optimized Cyber Twin (CT) based solution for task offloading with reduced offload latency in MFC. The proposed solution is built on the CT based Fog Computing (CyTFC) system where the computation node predicts server state with a DRL-based Multi-Agent Actor-Critic (MA3C) framework and provides the training dataset for task offloading prediction with a reward function. Furthermore, a delay-efficient offloading scheme is also proposed at the CT fog computing nodes. The Lyapunov optimization function was also modified to lower long-term migration costs. Finally, a deep reinforcement learning method called Multi-Agent Actor-Critic (MA3C) is suggested to solve the multi-objective dynamic optimization issue. The proposed CT based solution exceeds existing state-of-the-art benchmarks with network fairness up to 92% and reduces offloading time, offloading failure rate, and task migration cost by 30%, according to the performance assessment findings from the thorough simulation.