Lossless compression of voiceband data signals

Today's public switched telephone networks (PSTN) are being upgraded to use digital transmission and switching. The main disadvantage of digital transmission is the higher bandwidth required. To alleviate this problem, much research has been devoted to the compression of digitized voice signals. One problem with voiceband data traffic over a digital PSTN is the seemingly inefficient use of bandwidth. Another problem is that voice compression techniques tend to have adverse effects on non-voice signals such as modems and faxes.

Literature search has turned up very little research in the area of actual compression of voiceband data signals. The most comprehensive works seems to have been done at the University of Salt Lake City in the late 1980s. They examined the use of Vector Quantization for compression of voiceband data (modem) signals. They only examined modem signals up to 4800bit/s. Since the time of their research modem speeds have increased to 28.8kbit/s and above.

This thesis experimentally examines whether or not it is possible to significantly compress digitized modem signals that are transmitted over the public telephone network in order to save bandwidth. The use of lossless compression schemes such as dictionary (Ziv-Lempel) and statistical (arithmetic and Huffman) compression were examined. These coders are able to reduce the bandwidth required for digitized modem signals by 5% to 25%. However, lossless coders have a variable bit-rate output and would be difficult to integrate into the public telephone network since it uses fixed bit-rate channels.

Any attempt to compress modem signals would require many resources and excessive time to implement. A better solution is to adopt a completely digital network end-to-end, e.g. ISDN. (Abstract shortened by UMI.)