Title :
More on the error recovery for variable-length codes
Author :
Swaszek, Peter F. ; DiCicco, P.
Author_Institution :
Dept. of Electr. & Comput. Eng., Rhode Island Univ., Kingston, RI, USA
fDate :
11/1/1995 12:00:00 AM
Abstract :
Variable-length codes (e.g. Huffman codes) are commonly employed to minimize the average codeword length for noiseless encoding of discrete sources. Upon transmission over noisy channels, conflicting views note that such codes “tend to be self-synchronizing” and suffer from the “catastrophic effect of the error´s propagation”. Maxted and Robinson (1985) used a state model to describe the error recovery of the decoder. The present paper extends their results in several ways
Keywords :
Huffman codes; channel coding; decoding; error correction codes; memoryless systems; source coding; telecommunication channels; variable length codes; Huffman codes; average codeword length; decoder; discrete sources; error propagation; error recovery; noiseless encoding; noisy channels; self-synchronizing; state model; variable-length codes; Analytical models; Binary codes; Computer errors; Decoding; Delay; Entropy; Error correction codes; Probability; Source coding; Testing;
Journal_Title :
Information Theory, IEEE Transactions on