With the increase in the number of vehicles and intelligent applications in the IoV, computing intensive tasks have proliferated, and the traditional architecture is difficult to meet user needs. To solve the problems of insufficient and uneven allocation of computing resources in the IoV, insatiable application delay requirements, and high task energy consumption costs, this paper combined MEC and SDN to design an efficient task offloading scheme in the network of vehicles architecture from macro station to MEC server to vehicle, and proposed an improved low-complexity non-dominated sorting genetic algorithm to optimize the task offloading cost and the load balancing rate of MEC server. The experimental simulation results show that the proposed scheme compared with random offloading, NO-MEC offloading, NO-I offloading, traditional NSGA, NSGA-Ⅱ offloading, GA offloading, Q-learning offloading, DQN offloading, has lower offloading cost, better load balancing rate, and approximately the highest system utility, which brings better network services to vehicle users in the IoV.

[2023-01-31] Accepted Paper /doc/2022.10.0630/v1

[2023-06-05] Printed Article /doc/2022.10.0630/v2