Enhanced feature detection and tracking algorithm for UWB-OFDM SAR navigation

There is a great need to develop non-GPS based methods for position and navigation in situations where GPS is not available. This paper presents an improved feature extraction and tracking algorithm for a novel navigation sensor, an Ultra-Wideband Orthogonal Frequency Division Multiplexed radar. The radar is onboard an aerial vehicle and collects data periodically in a side-looking stripmap Synthetic Aperture Radar configuration flying over unknown terrain. An M/N detector is first used to extract range/Doppler observations from SAR features in the terrain. The observations are then associated with existing tracks using Global Nearest Neighbor. The associated measurements are then used as updates to a previously developed EKF which combines the radar observation tracks with INS measurements to compute an optimal aircraft position estimate. The filter performance is evaluated and compared with that of a previously developed threshold-based method using simulations. The algorithm performance dependence on the quality of INS measurements are also evaluated based on INS error models of commercial-grade, tactical-grade, and navigation-grade INS systems.