Lossless data compression with error correcting codes

Existing zero-error variable-length data compression algorithms suffer from sensitivity to transmission errors and error propagation across packets when used in packet-based data transmission through noisy channels. We propose a new approach to lossless data compression based on error correcting codes and the block-sorting transform. The proposed data compression scheme is compared with the universal algorithm for a four-state Markov chain in comparison with LZ (gzip) and a conventional BWT-based compressor (bzip). Although the comparison is quite encouraging, the main motivation and advantages of our scheme over the conventional separation-based approach accrue in the joint source/channel setting.