Mobility-Aware Multi-Objective Offloading Optimization in MEC and Vehicular-Fog Systems: A Waited-Ratio Based TD3 Approach

Multi-access Edge Computing (MEC) and Vehicular-Fogs (VFs) are placed nearer to user equipment (UE), reducing propagation latency compared to traditional cloud-based systems and ensuring a high standard of Quality of Service (QoS). Nevertheless, MEC sites can become congested and overloaded during peak traffic periods, such as concerts or sporting events. To address this, offloading techniques can shift intensive computational tasks from devices with limited resources to those with greater capacity, enhancing task performance and thereby increasing battery longevity. This study investigates the offloading within a two-tier framework of MEC and VF, focusing on the offloading of MEC to VF. Maintaining QoS is challenging due to the instability of fog networks caused by high-speed vehicle movement, which disrupts both vehicle-to-vehicle and vehicle-to-infrastructure communications. To mitigate this, we analyze vehicle mobility using a Gauss-Markov Mobility (GMM) model. Our main goal is to reduce the average system cost by optimizing both latency and energy consumption while accounting for vehicle mobility. We approach this challenge as a multi-objective optimization problem and develop a reinforcement learning environment. Additionally, we propose an algorithm based on imitation learning called Weighted-Ratio Based TD3 (WRTD3), an enhancement of the TD3 algorithm, to effectively manage these complexities.