A Subdivision-Fusion Algorithm for Radar Range Super-Resolution in Gapped Bands

A system of co-located narrow-band radars operating in disjoint frequency bands may be viewed as a gapped-band system, which can then be cast into a compressed sensing (CS) formulation for improved scene resolution. However, such a gap increases the coherence of the sensing matrix, causing existing CS algorithms to fail. A subdivision-fusion algorithm is discussed to tackle this problem, along with its super-resolution capabilities for different band-gaps. Test results on real radar data are presented. Effects of horizontal random projection of such a gapped-sensing matrix are also discussed.