An Efficient Offloading Algorithm Based on Support Vector Machine for Mobile Edge Computing in Vehicular Networks

Abstract

In vehicular networks, Mobile Edge Computing (MEC) is applied to meet the offloading demand from vehicles. However, the mobility of vehicles may increase the offloading delay and even reduce the success rate of offloading, because vehicles may access another road side unit (RSU) before finishing offloading. Therefore, an offloading algorithm with low time complexity is required to make the offloading decision quickly. In this paper, we put forward an efficient offloading algorithm based on Support Vector Machine (SVMO) to satisfy the fast offloading demand in vehicular networks. The algorithm can segment a huge task into several sub-tasks through a weight allocation method according to available resources of MEC servers. Then each sub-task is decided whether it should be offloaded or executed locally based on SVMs. As the vehicle moves through several MEC servers, sub-tasks are allocated to them by order if they are offloaded. Each server ensures the sub-task can be processed and returned in time. Our proposed algorithm generate training data through Decision Tree. The simulation results show that the SVMO algorithm has a high decision accuracy, converges faster than other algorithms and has a small response time.