Variational Inference-Based Positioning with Nondeterministic Measurement Accuracies and Reference Location Errors
Variational Inference-Based Positioning with Nondeterministic Measurement Accuracies and Reference Location Errors
复制标题
DOI：
10.1109/tmc.2016.2640294
复制
发表时间：
2017-10
影响因子：
7.9
通讯作者：
Bingpeng Zhou;Qingchun Chen;H. Wymeersch;P. Xiao;Lian Zhao
中科院分区：
文献类型：
--
作者：
Bingpeng Zhou;Qingchun Chen;H. Wymeersch;P. Xiao;Lian Zhao
协作网络定位是无线传感器网络中的一个重要问题，相邻传感器节点之间可以相互帮助进行定位。但由于无线传播环境的动态变化和周围环境的不同，不同网络节点的测量精度是不同的，并且随着时间的推移而变化。本文考虑了测量精度和参考节点位置的不确定性，以考虑不同的环境条件和初始节点定位误差对协作网络定位的影响。提出了一种基于均值场变分推理的协作网络定位算法。VIP算法的机制，收敛特性，实现复杂性，和并行实现结构的介绍表明，VIP算法提供了一个有效的机制，合并和共享所有网络节点之间的定位信息，以提高定位性能。最后，给出了一个简洁的Cramer-Rao下界，揭示了定位误差的传播规律。发现定位误差传播原理类似于电路理论中的欧姆定律，这为研究测量精度、参考节点定位误差和参考节点数目对协作网络定位性能的影响提供了新的视角。
Cooperative network localization plays an important role in wireless sensor network (WSN), wherein neighboring sensor nodes will help each other to calibrate their locations. However, due to the dynamic wireless propagation environment and different surroundings, the measurement accuracy at different network nodes is different and varies over time. In this paper, the uncertainties in both measurement accuracy and reference node locations are considered to account for the impact of different surrounding environments and the initial node location errors on the cooperative network localization. A mean-field variational inference-based positioning (VIP) algorithm is proposed for cooperative network localization. The mechanism of the proposed VIP algorithm, the convergence properties, implementation complexity, and the parallel implementation structure are presented to show that the VIP algorithm provides an effective mechanism to incorporate and share the localization information among all network nodes for an improved localization performance. Finally, a concise Cramer-Rao lower bound (CRLB) is derived to reveal the principle of localization error propagation. It is disclosed that the localization error propagation principle is similar to the Ohm's Law in circuit theory, which provides a new insight into the impact of the measurement accuracy, the reference node location errors and the number of reference nodes on the cooperative network localization performance.