Task-driven charger placement and power allocation for wireless sensor networks
Task-driven charger placement and power allocation for wireless sensor networks
复制标题
DOI：
10.1016/j.adhoc.2021.102556
复制
发表时间：
2021-08
期刊：
影响因子：
4.8
通讯作者：
Xingjian Ding;Jianxiong Guo;Yongcai Wang;Deying Li;Weili Wu
中科院分区：
文献类型：
--
作者：
Xingjian Ding;Jianxiong Guo;Yongcai Wang;Deying Li;Weili Wu
部署无线充电器是为电池驱动的传感器节点提供持续能量供应的一种很有前途的方法，近年来受到越来越多的关注。现有的工作主要集中在研究充电效用最大化问题，即，他们的目标是提供尽可能多的能量给所有传感器节点与一定数量的无线充电器。然而，这些工作忽略了传感器节点的任务执行功能，提供更多的能量给传感器节点并不能保证任务能够更好地执行。因此，在本文中，我们考虑了部署无线充电器的更实际的问题，我们的目标是在有限的部署成本预算下最大化总的任务效用。我们将问题转化为一个混合整数非线性规划问题，并证明了其NP-困难性。为了解决这个问题，我们将其分为两个子问题，其中第一个子问题是关于传感器节点的功率分配，第二子问题是关于无线充电器的最佳放置。我们为每个子问题设计了一个近似算法，并对算法进行了理论分析。通过相应地解决这两个子问题，我们得到了一个可行的解决方案，该解决方案对原始问题具有1 2（1 - 1 e）的性能保证。最后，我们进行了大量的模拟，以验证我们的算法的性能。结果表明，我们的设计显着优于基线算法，这证明了我们的算法的有效性。
Deploying wireless chargers is a promising way to provide continuous energy supply for battery-driven sensor nodes, and has attracted more and more attention recently. Existing works mainly focus on studying the charging utility maximization problem, that is, they aim to supply as much power as possible to all sensor nodes with a certain number of wireless chargers. However, these works ignore the task performing function of sensor nodes, supplying more power to sensor nodes cannot guarantee that tasks can be performed better. In this paper, therefore, we consider a more practical issue of deploying wireless chargers, where our objective is to maximize the total achieved task utility with a limited deployment cost budget. We formulate our problem as a mixed integer non-linear programming problem, and prove its NP-Hardness. To address this problem, we split it into two sub-problems, where the first sub-problem is about power allocation for sensor nodes, and the second sub-problem is about optimal wireless charger placement. We design an approximation algorithm for each sub-problem, and give theoretical analyses of our algorithms. By solving the two sub-problems accordingly, we get a feasible solution that has a 1 2 (1− 1∕ e) performance guarantee for the original problem. Finally, we conduct extensive simulations to validate the performance of our algorithms. The results show that our designs significantly outperform the baseline algorithms, which demonstrates the effectiveness of our algorithms.