Portal de Periódicos da CAPES

Quantizing Noise of ΔM/PCM Encoders

1973; Institute of Electrical and Electronics Engineers; Volume: 52; Issue: 2 Linguagem: Inglês

10.1002/j.1538-7305.1973.tb01958.x

2376-7154

D.J. Goodman, L.J. Greenstein,

Algorithms and Data Compression

We consider the pulse-code-modulation encoder that contains a delta modulator for analog-to-digital conversion, and a finite impulse response digital filter that suppresses high-frequency components of the delta modulation signal. A PCM word generator produces fixed-length binary code words by rounding and amplitude limiting the filter output samples. The quantizing noise of the resulting PCM signal has four components: delta modulation slope overload noise, filtered delta modulation granular noise, amplitude overload noise, and word generator roundoff noise. We analyze the total quantizing noise for the case where the encoder input is a Gaussian random process and the digital filter impulse response is uniform (all coefficients equal). Such filters possess important implementation advantages and appear to be near optimal with respect to signal-to-noise performance. Our analysis results in curves which show the relationship of signal-to-noise ratio to filter order, delta modulation sampling rate, and PCM word length.