Energy-Efficient Resource Management for Mobile Edge Computing-Enabled Roadside Units in Multi-Vehicle Networks

With the advancement of vehicular networking technology, communication between vehicles, and between vehicles and cloudlets, is becoming increasingly frequent, leading to a growing demand for computing resources. This growing demand necessitates more robust and efficient computing solutions to handling the data exchange and processing requirements. Mobile edge computing (MEC) addresses computing demands by leveraging edge resources. In practice, numerous parameter constraints, such as task volumes and available resources, render optimal resource management challenging. This paper presents a vehicular networking communication scenario involving an MEC-enabled roadside unit and multiple vehicles. We propose a new method that jointly optimizes task offloading decisions along with power and bandwidth allocation, aiming to minimize system energy consumption. Given the non-convexity of the original problem, characterized by the complexity and interdependence of multiple optimization variables, we adopt a strategic approach to decouple it into two sub-problems. The problem can be solved using deep learning and subgradient methods separately. Finally, a refined solution can be obtained through iterative solving with the block coordinate descent (BCD) method. Simulations provide compelling evidence that our scheme significantly reduces system energy consumption, outperforming benchmarks and showcasing its superiority.