RDRTec and our partner NRL (Naval Research Lab) along with support from Telephonics propose to prototype and demonstrate a novel, efficient, two-prong Dwell Evaluation Expert System (DEES) plus compressive sensing approach to reducing the probability of an adversary recognizing that a sensor is executing ISAR dwells while maintaining maritime classification performance. The first path is to refactor imaging and classification algorithms to produce data quality and classification accuracy assessments at fine real-time intervals throughout the algorithm which will be assessed by DEES, allowing for exiting of the ISAR session when point of diminishing returns related to the ability to classify maritime targets has been reached. The second path is to reduce the total illumination time and staring characteristics of an ISAR session by segmenting the session into sub-sessions and using compressive sensing techniques to reform imagery. The proposed approach addresses both exiting sessions quickly when data quality is poor, and when data quality is good, automatically exiting the session when sufficient data has been collected to support target classification. To ensure that the technique can be transitioned to the widest possible set of sensors, we will restrict this effort to making no change to the characteristics of existing ISAR sessions
Benefit: RDRTec and our partner NRL with support from Telephonics propose to prototype and demonstrate a novel, efficient, two-prong Dwell Evaluation Expert System (DEES) plus compressive sensing approach to reducing the probability of an adversary recognizing ISAR dwells while maintaining maritime classification performance. We will refactor MCAs imaging and classification algorithms to produce data quality and classification accuracy assessments at fine real-time intervals throughout the algorithms for assessment by DEES, allowing for exiting ISAR sessions when further data will no longer improve maritime target classification. The second path is to reduce the total illumination time and staring characteristics of an ISAR session by segmenting the session (under-sample) into sub-sessions and using compressive sensing techniques to reform imagery. The proposed approach addresses both exiting sessions quickly when data quality is poor, and when data quality is good, automatically exiting the session when sufficient data has been collected to support target classification. To ensure that the technique can be transitioned to the widest possible set of sensors, we will restrict this effort to making no change to the characteristics of existing ISAR sessions other than turning those sessions on and off (e.g. no new waveforms or added pulse-to-pulse diversity).
Keywords: adversary, adversary, DEES, ISAR, Compressive sensing, MCA, Maritime Classification
