Simultaneous Target and Multipath Positioning

This paper addresses the problem of target geo-localization in complex multipath environments such as indoor and urban settings. In both radar applications (where targets are non-cooperative) and navigation in GPS-denied areas using RF signals (where the subject is cooperative), multipath propagation is a well-known cause of large geo-location errors except in rare cases when a very accurate channel model is available. This work addresses more typical situations of uncertain channel properties by jointly estimating target position and multipath parameters. The proposed Simultaneous Target and Multipath Positioning (STAMP) approach involves an application of multi-scan multi-hypothesis data association to approximate recursive Bayesian estimates of both moving target location as well as specular reflector and point scatterer locations. STAMP achieves joint estimation by exploiting the different dynamics of targets (e.g., people moving) versus channel parameters (e.g., fixed wall locations). Algorithm performance is evaluated in simulation for radar localization of a non-cooperative target in an uncertain urban multipath environment. In addition, the successful demonstration of STAMP geolocation using real wideband microwave data collected in an actual building foyer with unknown floor plan is discussed. Finally, the issue of identifiability of both target and multipath parameters is explored via analysis of the Cramer-Rao Lower Bound (CRLB) on joint estimation of target and multipath parameters in both line-of-sight and non-line-of-sight scenarios.