Lognormal Mixture Cramer-Rao Lower Bound for Localization

Abstract

In received signal strength (RSS) based localization problems, the accuracy of the position information obtained is closely associated with the RSS model used. Therefore, positioning success can be improved with a more accurate RSS model. In this study, to analyze the effect of RSS model in localization performance, lognormal mixture shadowing model is used that provides a more accurate RSS model than the classical lognormal shadowing model. For the corresponding mixture model, a tight upper bound for Cramer-Rao lower bound (CRLB) based on Jensen's inequality is derived. The obtained expressions are used in CRLB analyses that are employed in determination of the best estimates in localization. The improved localization accuracy with lognormal mixture shadowing model is demonstrated by means of examining various numerical analyses.