02.10.2019 Views

UploadFile_6417

Create successful ePaper yourself

Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.

562 Chapter 10 ROUND-OFF EFFECTS IN DIGITAL FILTERS<br />

Distribution of e 1<br />

Distribution of e 2<br />

4128<br />

3128<br />

2128<br />

1128<br />

4128<br />

3128<br />

2128<br />

1128<br />

SAMPLE SIZE N = 500000<br />

ROUNDED TO B = 6 BITS<br />

MEAN = 1.7026e–006<br />

MIN PROB BAR HEIGHT = 0<br />

MAX PROB BAR HEIGHT = 0.017902<br />

SIGMA = 0.0045105<br />

0<br />

−0.5 −0.375 −0.25 −0.125 0 0.125 0.25 0.375 0.5<br />

Normalized Error e 1<br />

SAMPLE SIZE N = 500000<br />

ROUNDED TO B = 6 BITS<br />

MEAN = 1.6953e–006<br />

MIN PROB BAR HEIGHT = 0<br />

MAX PROB BAR HEIGHT = 0.017606<br />

SIGMA = 0.0031059<br />

0<br />

−0.5 −0.375 −0.25 −0.125 0 0.125 0.25 0.375 0.5<br />

Normalized Error e 2<br />

FIGURE 10.14 Multiplication quantization error distribution for the sinusoidal<br />

signal in Example 10.9, B =6bits<br />

The plots of the resulting histogram are shown in Figure 10.14. For the sinusoidal<br />

signal, when B =6bits, the error samples are not uniformly distributed<br />

and the samples are not independent. The means of e(n) and [e(n)+e(n−1)]/2<br />

are small. Their standard deviations are 0.0045105 and 0.0031059, which do<br />

not agree with (10.10). The corresponding plots for B =12bits are shown in<br />

Figure 10.15 from which we observe that the quantization error sequence appears<br />

to satisfy the model assumptions for B ≥ 12 bits. The means of e(n) and<br />

[e(n)+e(n − 1)]/2 are very small, and their standard deviations agree closely<br />

with (10.10).<br />

□<br />

□ EXAMPLE 10.10 Let x(n) beanindependent and identically distributed random sequence whose<br />

samples are uniformly distributed over the [−1, 1] interval. Using 500,000 samples<br />

to minimize any statistical variations, analyze normalized errors.<br />

Solution<br />

The following MATLAB script computes the distributions for B =6bits.<br />

clear; close all;<br />

% Example parameters<br />

B = 6; N = 500000; xn = (2*rand(1,N)-1); bM = 7; c = 1/sqrt(2);<br />

% Signal and Coefficient Quantization<br />

xq = (round(xn*(2^B)))/(2^B); c = (round(c*(2^B)))/(2^B);<br />

cxq = c*xq;<br />

% Multiplication of constant and signal<br />

Copyright 2010 Cengage Learning. All Rights Reserved. May not be copied, scanned, or duplicated, in whole or in part. Due to electronic rights, some third party content may be suppressed from the eBook and/or eChapter(s).<br />

Editorial review has deemed that any suppressed content does not materially affect the overall learning experience. Cengage Learning reserves the right to remove additional content at any time if subsequent rights restrictions require it.

Hooray! Your file is uploaded and ready to be published.

Saved successfully!

Ooh no, something went wrong!