Cooperative Joint Synchronization and Localization in Wireless Sensor Networks
Cooperative Joint Synchronization and Localization in Wireless Sensor Networks
复制标题
DOI：
10.1109/tsp.2015.2430842
复制
发表时间：
2015-07-15
影响因子：
5.4
通讯作者：
Buehrer, R. Michael
中科院分区：
文献类型：
--
作者：
Vaghefi, Reza Monir;Buehrer, R. Michael
在本文中，研究了使用异步到达时间测量的协作传感器定位。众所周知，使用基于时间的测距或伪测距方法的无线网络中的定位性能很大程度上受到估计中涉及的节点之间的定时同步的准确性的影响。通常，原始估计问题被分解为两个子问题，即同步问题和定位问题，即所谓的两步方法。然而，在本文中，考虑并检验了联合同步和定位问题在协作网络中的使用。讨论了源节点之间的协作消除了对高锚节点密度的需要并显着提高了定位性能。此外，还导出了 Cramer-Rao 下界 (CRLB) 和最大似然 (ML) 估计量。结果表明，ML 估计器是高度非线性和非凸的，因此必须使用计算复杂的算法来求解。为了降低估计的复杂度，通过松弛原始非凸ML问题，开发了一种新颖的半定规划（SDP）松弛方法，从而将估计问题重新表述为凸问题。通过计算机模拟表明，所提出的 SDP 方法的性能几乎等于 ML 估计器的性能。该方法也适用于非合作情况，发现它的性能优于先前提出的次优估计器。最后，对所考虑的估计器进行复杂性分析。
In this paper, cooperative sensor localization using asynchronous time-of-arrival measurements is investigated. It is well known that localization performance in wireless networks using time-based ranging or pseudo-ranging methods is greatly affected by the accuracy of the timing synchronization between the nodes involved in the estimation. Commonly, the original estimation problem is broken down into two subproblems, the synchronization problem and the localization problem, in what is known as a two-step approach. However, in this paper, the joint synchronization and localization problem is considered and examined for use in cooperative networks. It is discussed that the cooperation between the source nodes eliminates the need for high anchor node densities and improves localization performance significantly. Furthermore, the Cramer-Rao lower bounds (CRLB) and the maximum likelihood (ML) estimator are derived. It is shown that the ML estimator is highly nonlinear and nonconvex and must, therefore, be solved by using computationally complex algorithms. In order to reduce the complexity of the estimation, a novel semidefinite programming (SDP) relaxation method is developed by relaxing the original nonconvex ML problem, in such a way as to reformulate the estimation problem as a convex problem. The performance of the proposed SDP method is shown through computer simulations to nearly equal that of the ML estimator. The approach is also applied to the noncooperative case where it is found to be superior in performance than the previously proposed suboptimal estimators. Finally, complexity analyses are included for the estimators under consideration.