Versatile algorithms for accurate indoor geolocation

Abstract

This paper tackles the challenging problem of accurate indoor geolocation for UWB systems in hazardous multipath environments through three versatile super resolution techniques: time domain MUSIC (Multiple Signal Classification), frequency domain MUSIC algorithms and frequency domain EV (Eigen Value) method. The resultant pseudo-spectrums generated by these algorithms can be used for positioning in both LOS (through accurate time delay estimation) and Non-LOS (by using the resolved multipath profile as a fingerprint for location based fingerprinting techniques) environments. The impact of Eigen value de-weighting on multipath resolvability is demonstrated and its practical implications highlighted. The superior noise immunity of the time domain MUSIC algorithm is verified and introduced as a prime candidate for noisy Non-LOS environments. Finally the spectral leakage effect of the time domain MUSIC algorithm is used to our advantage to generate variants of the algorithm to provide better positioning accuracy for relatively low bandwidth conditions.