The evolution of lossless data compression techniques

Lossless data compression has suddenly exploded into a $300 million market. Commercial examples are file compression software packages than can more than double the storage in disc files while the new CCITT V.42bis standard can double the transmission rates in modems. Applications will soon spread into digital image compression (HDTV), teleconferencing, wide area networks (WAN, ISDN), digital audio tapes (DAT) and brain-like databases (Autosopher). There are five competing methods each championed by different vendors and the battle for the best algorithm is far from being settled. The old Huffman code was long a favorite of academics and mathematicians. It may be used as an addition to other compression codes but is not commercially viable by itself. The original Ziv-Lempel (1977) codes have a temporary headstart because of heavy promotions by Stac Electronics. These codes already deliver slow file compression software or fast hardware compression using plug-in modules. These codes will prove to be inferior in the long run. The Ziv-Lempel-Welch (LZW) codes are now the most commonly used compression codes. They may be superseded soon by combined fixed and dynamic learning networks. The British Telecom Lempel-Ziv (BTLZ) code is the state of the art for dynamic libraries. It is used in the V.42bis compression modems. The fixed library Autosophy networks offer the highest compression ratio and additional data encryption. They do however require prior knowledge of the kind of data or language being compressed. These networks may operate at extreme speed and without the danger of error propagation.