Maximum Averaged Likelihood Estimation Tree for Anchor-Less Localization Exploiting IR-UWB Multipaths

This paper presents an algorithm that enables anchor-less localization for pairs of static Impulse Radio - Ultra Wide Band (IR-UWB) devices in a room. The estimation progresses through a tree, as the most significant received paths are successively considered. Unlikely path parameters estimates are progressively rejected, relying on cross-correlation (i.e. with respect to predicted waveforms), as well as on a Maximum Averaged Likelihood (MAL) criterion. The latter is based on the measurements of Angle of Incidence (AoI) and Time of Arrival (ToA) for each detected multipath component, in addition to one range measurement. Once intrinsic mirror and transmitter/receiver ambiguities are solved, in the lack of antennas array and reference landmarks, a positioning error lower than 0:6 m and a room dimension error lower than 2% are obtained for 95% of estimates.