A Dynamic Task Offloading Scheme Based on Location Forecasting for Mobile Intelligent Vehicles

Intelligent vehicles integrate offloading into traditional transportation systems and migrate tasks from resource-constrained vehicles to other vehicles or infrastructures for collaborative tasks through this technology, effectively improving the overall utility and reducing the task computing time. However, few of the existing task offloading schemes have effectively considered the mobility characteristics of vehicles, which leads to frequent task offloading failures. In this paper, we take the future motion trajectory, resource load, and offloading delay of vehicles into account and propose a dyFnamic task offloading scheme based on location forecasting. The scheme that combines the location prediction model can effectively improve the task offloading success rate while reducing the offloading delay. To verify the effectiveness of our proposed scheme, we conduct simulations and real-world scenario experiments. The results of the simulations and real-world scenario experiments show that our proposed scheme outperforms other comparative baseline schemes in both offloading delay and offloading success rate dimensions. In detail, our scheme can reduce the offloading delay by up to 56.68% and 88.63% compared to the other two schemes, and improve the offloading success rate by up to 14% and 16.5% compared to the other two schemes.