A Novel, Flexible, and Comprehensive System for Mission Prioritized Lossless Data Compression

ABSTRACT: The test and evaluation (T & E) mission in complex test facilities results in large amounts of data. Moreover, the data may have different characteristics, including images from particle image velocimetry data, air flow data, etc. Finally, some of the data may come from a network of similar sensors. We propose a novel, flexible, and comprehensive system for mission prioritized lossless data compression. First, we propose to apply our latest compressed sensing (CS) algorithm known as singular value decomposition-QR (SVD-QR) to jointly compress sensor data in a sensor network. We have successfully applied SVD-QR to actual radar sensor network data with 30 sensors from the Air Force and achieved a compression ratio of 192 without loss of information. Second, we propose a high performance CS algorithm that can efficiently compress sensor data that can be modeled as auto-regressive hidden Markov model (AR-HMM). Acoustic and radio frequency (RF) signals can be characterized by AR-HMM. Third, for isolated sensors such as particle image velocimetry and other pressure and force sensors, we propose to apply a new algorithm called CS-SVD (compressed sensing - singular value decomposition) to perform the compression. All of our algorithms have parameters that allow users to choose for different mission priorities. BENEFIT: The proposed technology will be useful for large data compression in test facilities such as military bases and NASA. Other applications include data compression for sensor networks, image compression for surveillance and reconnaissance operations, and also compression for commercial camcorder and digital cameras. We envision the market for the system developed will be 50 million dollars over the next decade.