A Sojourn Time based Algorithm for Vehicular Edge Task Offloading

Vehicular edge computing (VEC) involves deploying edge servers in vehicles moving along highways, to provide additional computing resources for low latency and high reliability in computational tasks. However, the dynamic nature of VEC environments, particularly regarding the task offloading of mobile vehicles acting as mobile edge servers, requires further research. To address this challenge, we propose a novel delay optimization algorithm based on deep reinforcement learning in VEC scenarios. It considers intelligent vehicles as mobile service nodes, collaborating with fixed roadside unit service nodes to form an integrated task offloading model. This model facilitates computational task offloading through vehicle-to-vehicle (V2V) and vehicle-to-infrastructure (V2I) communications. It accounts for the sojourn time routing planning of vehicles and supports divisible tasks, thereby improving offloading efficiency. Then, an Asynchronous Advantage Actor-Critic (A3C) algorithm is used to achieve an effective computational offloading strategy. Simulation results demonstrate that the algorithm significantly outperforms existing methods in terms of the total system delay, while providing a more efficient and effective solution for task offloading in VEC scenarios.