Adaptive Compression of Telemetry Data using Application Specific Dictionaries

Vadum proposes to create a novel telemetry data compression/decompression system, SCoCD (Sparse Compression on Complete Dictionaries) with very low latency, very high compression rate, and very low distortion. The system will allow lossless compression, and can be tuned for each type of data stream with particular emphasis on imagery. Vadum has demonstrated x59 compression of Aegis test-fire imagery, with a peak SNR of 45 dB, in less than 0.02 seconds computation time with a slow, high-level language implementation. Classical compression techniques achieve low-latency, high-compression rate constraints but fail to achieve low distortion constraint, particularly for imagery. Hardware-accelerated solutions may achieve low latency and low distortion, but lack compression performance when applied to video or images. None of the classic techniques can compress already encrypted data. SCoCD will fill the technology gap by integrating the latest advances of Compressive Sensing and Dictionary Learning.