Statistical Analysis of Telemetry Data for Compression
Abstract
Introduction: The problem of adaptive compression implies the need for operational analysis (classification) of the compressed data in order to select the most efficient compression algorithm and set its optimal parameters. Classifying data by functional features does not take into account the compression of the measurement information, so classification on the base of statistical properties should be considered more appropriate. Purpose: We estimate the prospects of compression algorithms which take into account statistical and autocorrelation properties of telemetry data. Results: Studying the distribution histograms of the raw data and its differential view has shown that a large dynamic range does not allow you to efficiently solve the problem of non-stationary data compression, but for stationary data you can predict a sufficiently high compression efficiency. The studies of one-dimensional and two-dimensional autocorrelation functions of various telemetry data types suggest that the most efficient compression algorithms can be those which take into account the correlations between references of a single telemetry frame and between different frames of a data flow. The studies have shown that the most promising algorithms represent a telemetry frame in a difference-bit form, taking into account the way in which data from separate sources merge into telemetry frames. We proposed basic principles of a compression method based on repeated delta encoding of the analyzed data elements which correspond to the first and second maximum of a one-dimensional autocorrelation function. For further development of telemetry data compression methods, we suggest to search for homogeneous structures inside a telemetry frame.Published
2017-02-20
How to Cite
Bogachev, I., Levenets, A., & Chye, E. (2017). Statistical Analysis of Telemetry Data for Compression. Information and Control Systems, (1), 11-16. https://doi.org/10.15217/issn1684-8853.2017.1.11
Issue
Section
Information processing and control