Adaptive Quantization with Prediction at the Receiver

Computer simulation has been made earlier on the adaptive quantizer at the transmitter with prediction at the receiver, to study the effects of different quantizer parameters on the SNR value of such a quantization scheme. To eliminate the need of a separate channel, the quantization step size was predicted earlier at the receiver. In this paper, the input source taken is a Gauss-Markov sequence and studies are made for different values of wordlength, range factors and for different input sequences. It is shown that for a resonably large wordlength, this type of quantization gives a better SNR value than most other existing quantization schemes.