Channel Noise - A Limiting Factor on the Performance of a Class of Adaptive Techniques

Abstract

The effects of channel noise on a class of adaptive sampling techniques based on the concept of removal of redundant data samples were investigated. Assuming a system of fixed bandwidth and fixed transmitter power, the channel noise forces the adaptive system to operate at a lower bit error probability than the equivalent PGM system since in the adaptive system each transmitted bit represents more information. This limitation was partially overcome by adding error criterion of unequal weighting to the data which enabled the system to operate at essentially the same bit error probability as the PCM system with a net coding efficiency greater than the coding efficiency of the error-correction code. Experimental results from subjective tests, and the RMS error demonstrate that a new error criterion must be developed for the class of adaptive techniques.