Narrative Compression Coding for a Channel with Errors.
Abstract
Data compression codes offer the possibility of improving the thruput of existing communication systems in the near term. This study was undertaken to determine if data compression codes could be utilized to provide message compression in a channel with up to a 0.10 bit error rate. The data compression capabilities of codes were investigated by estimating the average number of bits-per-character required to transmit narrative files. The performance of the codes in a channel with errors (a noisy channel) was investigated in terms of the average numbers of characters-decoded-in-error and of characters-printed-in-error-per-bit-error. Results were obtained by encoding four narrative files, which were resident on an IBM-PC and use a 58 character set. The study focused on Huffman codes and suffix/prefix comma-free codes. Other data compression codes, in particular, block codes and some simple variants of block codes, are briefly discussed to place the study results in context. Comma-free codes were found to have the most promising data compression because error propagation due to bit errors are limited to a few characters for these codes. A technique was found to identify a suffix/prefix comma-free code giving nearly the same data compressions as a Huffman code with much less error propagation than the Huffman codes. Greater data compression can be achieved through the use of this comma-free code word assignments based on conditioned probabilities of character occurrence. Keywords: Comma-free data, Data compression capabilities, and Huffman code.
Document Details
- Document Type
- Technical Report
- Publication Date
- Jan 01, 1988
- Accession Number
- ADA191577
Entities
People
- James W. Bond