Adaptive Task Scheduling via End-Edge-Cloud Cooperation in Vehicular Networks

Abstract

Service latency is one of the most crucial factors to be considered in vehicular networks, especially for safety-critical applications with complex tasks. This paper presents a three-layer service architecture in vehicular networks, including the cloud layer, the edge layer and the terminal layer, and the nodes on different layers have different capacities on task processing. Specifically, vehicles may generate a set of tasks, and each task may be composed of multiple subtasks, which may require different amount of computation and memory resources for processing, and the task is served only when all of its subtasks are completed. On this basis, we formulate an adaptive task scheduling (ATS) problem, with the objective of minimizing the overall service latency by best cooperating those heterogeneous nodes on the cloud, edge and terminal layers. Further, we propose a genetic algorithm GA_ATS to solve the problem. In particular, we design a real number vector representation for encoding solutions, a fitness function for solution evaluation, a set of crossover and mutation operations for offspring generation and a balanced greedy algorithm for fixing infeasible solutions. Finally, we build the simulation model and conduct a comprehensive performance evaluation. The result shows that our proposed algorithm can effectively improve the system performance in terms of minimizing service latency.