Digital Signal Processing Reference
In-Depth Information
13
Estimation of Correlation
Functions
Continuing the discussion on the specifics of processing nonuniformly quantized
signals, started in the previous chapter, the issues of quantized signal multipli-
cation are studied and put into focus here. They are of special interest in many
cases where two or more pseudo-randomly digitized signals are to be processed
together. An estimation of correlation functions represents the most typical case.
As a rule, numerous multiplications of the quantized signal sample values need to
be executed in order to perform signal correlation analysis. As these operations
are relatively time consuming, it is essential to rationalize them. That is especially
important in cases of special hardware development for correlation analysis. The
most suitable quantization method has to be selected first. Referring back to
Chapter 5, it is clear that application of pseudo-randomized quantizing should
be considered. However, it seems that multiplication of two pseudo-randomly
quantized signals is more complicated than multiplication of randomly or de-
terministically quantized signals. On the other hand, the properties of pseudo-
randomized quantizing are superior to the analogous properties of other quantizing
techniques.
Application of pseudo-randomized sampling for signal correlation analysis is
also discussed. It is shown that signal digitizing strongly impacts on the conditions
for the correlation analysis and that applying pseudo-randomized digitizing tech-
niques, if done skilfully, leads to certain desirable effects. Therefore it is worth
examining various processes related to this kind of signal correlation analysis
more closely.
