Wireless Sensor Localization Using Outlier Detection

Abstract

Received RF signal strength provides a cost-effective mechanism for distance estimation that is popularly used in wireless sensor networks (WSN) range-based localization. However, such range estimates are heavily affected by shadowing that can be caused by obstructions of the line-of-sight radio frequency (RF) signal. Multilateration using several range estimates obtained from RSSI can lead to large errors in sensor location if one or more of the range estimates are affected by shadowing. In this paper, we present a scheme that applies spatial correlation and a clustering mechanisms to remove the effect of shadowing in range based location estimation. We show the effectiveness of this scheme in minimizing the adverse effects of signal outliers using simulations.