Cooperative Task Allocation in Edge Computing Assisted Vehicular Crowdsensing

As a popular scenario of mobile crowdsensing, edge computing assisted vehicular crowdsensing (EVCS) encourages vehicles to participate in sensing data with the equipped devices. Due to the vehicular mobility, vehicles may dynamically enter and leave the coverage area of an edge node, leading to recurrent task allocations that consume excessive communication and computational resources. How to avoid recurring recruitment in task allocation is challenging. In this paper, we propose an optimization framework to facilitate task allocation by utilizing the cooperation between edge nodes. The proposed framework avoids complicated recruitment procedures while maximizing the connection time between the recruited vehicles and the edge node. Due to the NP-hardness of the formulated optimization problem, we design a reinforcement learning based algorithm to solve the problem with high accuracy and efficiency. Simulation results show the effectiveness of our proposed framework.

作为移动众包感知的一种常见场景，边缘计算辅助的车辆众包感知（EVCS）鼓励车辆使用其配备的设备参与感知数据。由于车辆的移动性，车辆可能会动态地进入和离开边缘节点的覆盖区域，导致反复的任务分配，这会消耗过多的通信和计算资源。如何在任务分配中避免重复招募是一项挑战。在本文中，我们提出了一个优化框架，通过利用边缘节点之间的协作来促进任务分配。所提出的框架避免了复杂的招募程序，同时最大化了被招募车辆与边缘节点之间的连接时间。由于所构建的优化问题具有NP难性，我们设计了一种基于强化学习的算法，以高精度和高效率解决该问题。仿真结果表明了我们所提出框架的有效性。