Compressive Sensing for DCGS-N Navy STTR FY2012.A Sol No.: Navy STTR FY2012.A Topic No.: N12A-T026 Topic Title: Compressive Sensing for DCGS-N Proposal No.: N12A-026-0199 Firm: Scientific Systems Company, Inc 500 West Cummings Park - Ste 3000 Woburn, Massachusetts 01801-6562 Contact: Les Novak Phone: (781) 933-5355 Web Site: www.ssci.com Abstract: Compressive sensing (CS) is a relatively new form of data sampling that shows promise to greatly reduce the amount of information required to acquire and reconstruct information from sources such as synthetic aperture radar (SAR) imagery, electro-optical (EO) imagery, and RF data. CS has interesting practical applications in processing/exploitation of imagery, signals, and other structured data. SSCI has applied CS-based processing to the formation of SAR imagery from phase-history data that has been degraded by interruptions in the SAR data collect. SSCI's CS-based image formation algorithm provides imagery having nearly optimum image quality (IQ) from a significantly reduced amount of data. SSCI has also demonstrated CS-based image formation of EO data, obtaining excellent EO imagery from highly compressed data. The IQ of CS-compressed SAR and EO imagery is sufficient for exploitation by DCGS-N image analysts. SAR exploitation modes include coherent/non-coherent change detection, automatic and analyst assisted target recognition, target tracking, etc.; EO exploitation modes include Wide Area Motion Imaging (WAMI), visual target ID, target tracking, EO change detection, etc. CS-based processing of the imagery permits the detection, classification and estimation functions with reduced dimensionality, providing increased operational rates over the original sources. Benefits: SAR, EO/WAMI, and RF systems employing Compressive Sensing to optimize compression and image formation are robust to jamming and minimally interfere with other (military or civilian) operations. Such systems are stealthier and more difficult to counter due to utilization of randomized transmit waveforms and sub-sampling of sensor data prior to distribution of exploited data products. CS-based processing systems may be designed to detect isolated sparse scatterers while reducing returns from high sea-state clutter by fitting sea-clutter signals in appropriate over-determined dictionaries, providing better detection performance against periscopes and small craft. Downlink requirements are minimized via reduced sampling afforded by CS. Return