Power-based localization in correlated log-normal fading aided by conditioning measurements

Abstract

The paper addresses the performance evaluation via the Cramer-Rao Lower Bound (CRLB) of power-based localization of a source in spatially-correlated log-normal propagation. The novel element is the inclusion and assessment of the impact of conditioning measurements on such performance. The proposed model parameterizes performance by both the sensor topology (density, positioning) producing the current measurements as well as by conditioning measurements (essentially, prior or training data) which reduce the statistical uncertainty in the model. Experimental results for indoor and outdoor environments are presented which quantify the expected localization accuracy, as well as identify practical issues to be further addressed. One main concern is the quantification of scaling on required sensor-network size for achieving a pre-specified localization accuracy.