GenAI-5GS Fusion Approach: A New Path for Enhancing Computation Offloading Efficiency of Complex Tasks in Internet of Vehicles

With the advancement of real-time communication and data interaction technologies in the Internet of Vehicles (IoV), the pivotal role of Mobile Edge Computing (MEC) in optimising vehicle service efficiency has become increasingly prominent. However, existing task offloading strategies still present significant room for improvement in handling complex tasks, delay control, and resource allocation for Roadside Units (RSUs). To reduce task offloading delay and enhance resource utilization, this paper proposes an offloading method integrating Generative Artificial Intelligence (GAI) with 5G Slicing (GenAI-5GS). First, a Graph-GAN-based vehicle trajectory prediction model is constructed. Operating with CPU/GPU utilization below 50% on resource-constrained RSUs, it enables precise prediction of vehicle movements to support proactive RSU deployment. Secondly, it innovatively integrates 5G slicing technology to dynamically allocate Resource Blocks (RB) within the dedicated 5.9GHz IoV band. By combining multiple Modulation and Coding Schemes (MCS) to accommodate tasks of varying priorities, which enables slicing-based management and flexible resource allocation for RSUs. Finally, research findings demonstrate that the proposed method enhances trajectory prediction accuracy and resource utilization while reducing offloading delay, all while safeguarding multi-dimensional Quality of Service (QoS) requirements. Furthermore, this approach significantly improves IoV complex task processing performance and system stability, delivering safer and more efficient service experiences for in-vehicle users, thereby possessing clear engineering deployment value.