Localization and its performance in wireless sensor networks

Abstract

Wireless sensor networks (WSNs) have been considered as a promising tool for many location dependent applications including area surveillance, search and rescue, mobile tracking and navigation, and so on. The geographic location information of nodes is requested by most applications of WSNs, and is also critical for improving the management of WSNs, such as topology control, packet routing, security, etc. Therefore, sensor localization, which is the process of identifying the locations of sensor nodes in a wireless sensor network (WSN), plays an important role in all those systems. Inspired by physical models and performance measures commonly adopted in the literature, this work studies sensor localization with emphasis on its performance, which is a challenging problem due to the extremely limited resources available at each low-cost sensor node and complicated mechanisms for achieving sensor localization in WSNs.

Based on the Cramer-Rao lower bound (CRLB), the first part of this work analyzes from a novel perspective the performance of single-hop sensor localization, where each sensor node directly measures distances or angles from anchor nodes at known locations to locate itself. That is, differently from the existing work, a random node geometry is considered by assuming that anchor locations are random variables, and the distribution, mean and standard deviation of a resulting random metric for the localization performance are then carefully investigated. Such an analysis provides a statistical view on the localization performance and assists to comprehensively understand sensor localization.

In multi-hop sensor localization, by contrast, not every sensor node is capable of directly measuring a sufficient number of distances from anchor nodes to locate itself, so that both localization and relevant performance analyses become enormously complicated. Therein, a crucial problem arises, viz. error propagation, which is still not well understood. The second patt of this work deals with the performance of range-based multi-hop sensor localization in one-dimensional WSNs based on a CRLB analysis, and analytically obtains the fundamental behavior of the localization performance and properties of error propagation. Flowing from this, the third part of this work focuses on the error propagation problem in the two-dimensional case, and identifies the relationship among the localization error statistics of a sensor node, the hop count from this sensor node to anchor nodes, the sensor density and the variance of distance measurement errors, which is greatly helpful for effectively designing and deploying WSNs. The final part of this work proposes a connectivity-based distance estimation method which helps to substantially improve the quality of connectivity-based sensor localization in static dense WSNs. This method employs a maximumlikelihood estimator (MLE) to estimate the distance between two adjacent nodes from their local connectivity information, namely the numbers of their common and non-common immediate neighbors. In addition, the performance of this method is thoroughly examined and its advantages are confirmed by testing it based on measurement data from a real environment as well as applying it in connectivity-based sensor localization. Show more