Strategic Sensing in Vehicular Networks Using Known Mobility Information
Strategic Sensing in Vehicular Networks Using Known Mobility Information
复制标题
DOI：
10.1109/tvt.2017.2774282
复制
发表时间：
2018-03-01
影响因子：
6.8
通讯作者：
Valaee, Shahrokh
中科院分区：
文献类型：
--
作者：
Alasmary, Waleed;Sadeghi, Hamed;Valaee, Shahrokh
本文研究了车载网络环境下的目标感知问题。首先，我们将目标定义为在道路上移动的车辆，传感器定义为路边摄像头。然后，我们研究了预测机动性对减少每个摄像机被激活的次数的影响，以保证目标的覆盖。我们使用机会调度器将感知问题描述为一个整数线性规划。然后，我们对该公式进行了扩展，提出了一种新的利用预测的移动性信息的覆盖策略调度器。然后，我们将该方法扩展到一个完全分布式的版本，并提出了一种通过在传感器之间交换消息的近似算法。利用马尔可夫可用度模型和跟驰可用度模型，通过仿真表明，利用预测的移动信息，激活的传感器数量显著减少。然后，我们分析了这两种调度算法，以量化在感知中利用移动性信息的增益。由于节点的易管理性，我们采用了独立的节点移动性模型。该分析由两个主要部分组成：根据感知成本和可行性概率计算移动性增益。我们的分析和仿真表明，在较高的可行性概率和较低的感知成本方面，感知目标的机动性得到了提高。
In this paper, we study the problem of sensing targets in the context of vehicular networks. First, we define targets to be the vehicles moving on the road and sensors to be the roadside cameras. Then, we study the effect of predicted mobility on reducing the number of times each camera is activated in order to guarantee the coverage of targets. We formulate the sensing problem as an integer linear program using an opportunistic scheduler. Afterward, we extend the formulation and propose a novel strategic scheduler for coverage, which utilizes the predicted mobility information. We then extend this method to a fully distributed version and propose an approximation algorithm by exchanging messages among the sensors. Using a Markovian and a car-following availability models, we show by simulations that the number of activated sensors is significantly reduced by utilizing predicted mobility information. After that, we analyze both schedulers to quantify the gain of utilizing mobility information in sensing. We adopt an independent node mobility model due to its tractability. The analysis is composed of two main components; calculation of mobility gain in terms of sensing cost and probability of feasibility. Our analysis and simulations demonstrate the gain of mobility in sensing targets in terms of higher probability of feasibility and lower sensing cost.