Strategy for Dependent Tasks Under Vehicle Formation Assistance

With wide application of the internet of vehicles and mobile edge computing, vehicles are equipped with more and more complex applications like image recognition and augmented reality. However, the limited computing resource of vehicles is an important challenge for these applications. The advent of vehicular edge computing alleviates the problem of limited vehicle computing resources, but an improper task offloading strategy can lead to more task latency and energy consumption. To address the task offloading problem of complex vehicular applications, the paper proposes a task offloading approach based on considerations of subtasks offloading sequences and subtask dependencies, which employs directed acyclic graphs to represent the subtasks of vehicular applications and introduces vehicle formations to reduce the computational burden of edge computing servers. We formalize the problem into a constrained bi-objective optimization problem and solve the optimization problem with an improved NSGA-II algorithm called IMNSGA-II. Simulation results demonstrate that the task offloading strategy based on the IMNSGA-II algorithm outperforms four existing task offloading strategies in terms of convergence speed and superior optimization capabilities.