Vehicle Edge Computing improves the Quality of Service of vehicular applications by offloading tasks to the VEC server. However, with the continuous development of computation-intensive vehicular applications, the limited resources of the VEC server will not be enough to support these applications. Volunteer Computing-Based Vehicular Ad-hoc Networking (VCBV) proposes a concept of using vehicles as resources, which is considered to be a promising solution. In this paper, we study the multi-dependent task offloading problem in order to quickly and economically handle the overload task of the requesting vehicle in VCBV. Considering both task execution delay and execution cost, we formulate the problem of offloading the multi-dependent tasks of requesting vehicles to minimize total task completion time and execution cost. Since the offloading problem is NP-hard, an improved multi-objective Ant Colony Optimization algorithm is proposed. Firstly, we use a density-based clustering algorithm to form volunteer alliances that can contribute idle resources. Secondly, based on the volunteer alliances and RSUs, we use Analytic Hierarchy Process (AHP) to initialize pheromone concentration to make better decisions. Then, we design the update strategy of the pheromone concentration and heuristic information. Finally, we introduce Pareto optimal relationship to evaluate the results. A large number of simulation results verify that our algorithm has better performance than other alternatives.

中文翻译：

---

基于蚁群优化算法的车辆辅助依赖任务卸载在车载边缘计算中的应用


车辆边缘计算通过将任务卸载到 VEC 服务器来提高车辆应用程序的服务质量。但是，随着计算密集型车辆应用程序的不断发展，VEC 服务器的有限资源将不足以支持这些应用程序。基于志愿计算的车辆自组网 （VCBV） 提出了一种以车辆为资源的概念，这被认为是一个很有前途的解决方案。在本文中，我们研究了多依赖任务卸载问题，以便快速、经济地处理 VCBV 中请求车辆的超载任务。综合考虑任务执行延迟和执行成本，我们制定了卸载请求车辆多依赖任务的问题，以最小化任务总完成时间和执行成本。由于卸载问题是 NP-hard 的，因此提出了一种改进的多目标蚁群优化算法。首先，我们使用基于密度的聚类算法来形成志愿者联盟，这些联盟可以贡献闲置资源。其次，基于志愿者联盟和 RSU，我们使用层次分析过程 （AHP） 来初始化信息素浓度，以做出更好的决策。然后，我们设计信息素浓度和启发式信息的更新策略。最后，我们引入 Pareto 最优关系来评估结果。大量的仿真结果验证了我们的算法比其他替代方案具有更好的性能。