Delay, Energy, and Outage Considerations in GenAI-Enhanced MEC-NOMA-Enabled Vehicular Networks

In recent years, vehicular networks have experienced substantial growth, transforming the landscape of the transportation industry. In particular, the issue of computing task offloading in vehicular networks is critically important. The two key considerations are offloading task delay and energy consumption. Thus, a multiple access (MA) scheme is necessary in vehicular networks to facilitate more efficient computation task offloading. Non-orthogonal multiple access (NOMA) is recognized as a promising candidate for multiple access in fifth-generation (5G) and beyond networks due to its potential to improve overall spectrum efficiency through superposition coding. Moreover, mobile edge computing (MEC) can minimize computation task delay and energy consumption by bringing computational resources closer to vehicles. Therefore, this paper examines the potential of employing mobile edge computing (MEC) with non-orthogonal multiple access (NOMA) in vehicular networks to reduce overall computation offloading task delay and energy consumption. Specifically, computation models for task delay and energy consumption are presented. Additionally, considering universal frequency reuse, we analyze interference-limited scenarios and allow interference from the transmissions of other vehicles and roadside units. The optimization problems for delay and energy are formulated based on the described models. Due to the non-convex nature of the optimization problem, they are solved numerically using the Python programming language. Furthermore, analytical expressions for outage probability are provided to evaluate the typical vehicle’s outage performance. Simulation results demonstrate the delay and energy performance of the MEC-NOMA-enabled vehicular networks compared to their OMA-based counterparts. The results indicate that MEC-NOMA achieves lower delay, reduced energy consumption, and improved outage performance compared to OMA-based vehicular networks.