Fast Two-Stage Lempel-Ziv Lossless Numeric Telemetry Data Compression Using a Neural Network Predictor
Rajasvaran Logeswaran (Multimedia University, Malaysia)
Abstract: Lempel-Ziv (LZ) is a popular lossless data compression algorithm that produces good compression performance, but suffers from relatively slow processing speed. This paper proposes an enhanced version of the Lempel-Ziv algorithm, through incorporation of a neural pre-processor in the popular predictor-encoder implementation. It is found that in addition to the known dramatic performance increase in compression ratio that multi-stage predictive techniques achieve, the results in this paper show that overall processing speed for the multi-stage scheme can increase by more than 15 times for lossless LZ compression of numeric telemetry data. The benefits of the proposed scheme may be expanded to other areas and applications.
Keywords: Lempel-Ziv, lossless compression, neural networks, prediction, two-stage
Categories: E.2, H.3.m