Adaptive Clutter Suppression for Multi-Channel Airborne Passive Radar Using Dictionary Learning

Target detection with passive airborne radar is a challenging application due to the heterogeneous nature of the ground clutter, its range dependence and the Doppler spread originating from the relative motion between transmitter and receiver. In this work, the clutter power spectra obtained by multi-channel processing of airborne passive radar measurements are exploited to take advantage of the sparsity of the data in the angle-Doppler domain. Sparse representation is then used to extract and suppress the clutter contribution, by employing a dictionary which is adaptively learned from consecutive targetfree examples of the data. The strategy has been applied on passive airborne radar data acquired during the APARTGAS trials in 2019 by injecting a synthetic land target with realistic speed and position. Results shows an improvement on the signal to interference ratio for a sensible range of target signal to noise ratio.