Optimal Sensor Placement for Target Localization in IoT Systems: A Cramér-Rao Bound Perspective
Sheng Xu, Linlong Wu, Kutluyıl Doğançay, M. R. Bhavani Shankar, K. C. Ho, Xinyu Wu
Abstract
Wireless sensor network (WSN) constitutes the backbone of the Internet of Things (IoT), where target localization is an essential function that enables various IoT applications. Apart from the observations and signal-to-noise ratio, the localization performance of a WSN is affected by the relative geometry between the sensors and the target. Placing the sensors at better positions can reduce the noise sensitivity to the target location estimate, making most IoT operations more effective. This article provides a comprehensive overview of the models, technologies, implementations and challenges of optimal sensor placement for IoT. Specifically, we first illustrate the influence of sensor placement on the target localization performance using the Cramér-Rao lower bound (CRLB) theory that defines the asymptotic performance of a location estimator. Then, several CRLB-based metrics to characterize the positioning accuracy are introduced and elaborated. Next, based on the desired performance criterion, major techniques for solving the optimal sensor placement problem are discussed, including the analytical, numerical and data-driven approaches. Finally, we demonstrate some practical implementations of IoT systems with sensor optimal placements for localization, followed by a summary of challenges for future research and conclusions.