-
Notifications
You must be signed in to change notification settings - Fork 1
/
Copy pathChap_8.html
1143 lines (845 loc) · 44.9 KB
/
Chap_8.html
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
37
38
39
40
41
42
43
44
45
46
47
48
49
50
51
52
53
54
55
56
57
58
59
60
61
62
63
64
65
66
67
68
69
70
71
72
73
74
75
76
77
78
79
80
81
82
83
84
85
86
87
88
89
90
91
92
93
94
95
96
97
98
99
100
101
102
103
104
105
106
107
108
109
110
111
112
113
114
115
116
117
118
119
120
121
122
123
124
125
126
127
128
129
130
131
132
133
134
135
136
137
138
139
140
141
142
143
144
145
146
147
148
149
150
151
152
153
154
155
156
157
158
159
160
161
162
163
164
165
166
167
168
169
170
171
172
173
174
175
176
177
178
179
180
181
182
183
184
185
186
187
188
189
190
191
192
193
194
195
196
197
198
199
200
201
202
203
204
205
206
207
208
209
210
211
212
213
214
215
216
217
218
219
220
221
222
223
224
225
226
227
228
229
230
231
232
233
234
235
236
237
238
239
240
241
242
243
244
245
246
247
248
249
250
251
252
253
254
255
256
257
258
259
260
261
262
263
264
265
266
267
268
269
270
271
272
273
274
275
276
277
278
279
280
281
282
283
284
285
286
287
288
289
290
291
292
293
294
295
296
297
298
299
300
301
302
303
304
305
306
307
308
309
310
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
334
335
336
337
338
339
340
341
342
343
344
345
346
347
348
349
350
351
352
353
354
355
356
357
358
359
360
361
362
363
364
365
366
367
368
369
370
371
372
373
374
375
376
377
378
379
380
381
382
383
384
385
386
387
388
389
390
391
392
393
394
395
396
397
398
399
400
401
402
403
404
405
406
407
408
409
410
411
412
413
414
415
416
417
418
419
420
421
422
423
424
425
426
427
428
429
430
431
432
433
434
435
436
437
438
439
440
441
442
443
444
445
446
447
448
449
450
451
452
453
454
455
456
457
458
459
460
461
462
463
464
465
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532
533
534
535
536
537
538
539
540
541
542
543
544
545
546
547
548
549
550
551
552
553
554
555
556
557
558
559
560
561
562
563
564
565
566
567
568
569
570
571
572
573
574
575
576
577
578
579
580
581
582
583
584
585
586
587
588
589
590
591
592
593
594
595
596
597
598
599
600
601
602
603
604
605
606
607
608
609
610
611
612
613
614
615
616
617
618
619
620
621
622
623
624
625
626
627
628
629
630
631
632
633
634
635
636
637
638
639
640
641
642
643
644
645
646
647
648
649
650
651
652
653
654
655
656
657
658
659
660
661
662
663
664
665
666
667
668
669
670
671
672
673
674
675
676
677
678
679
680
681
682
683
684
685
686
687
688
689
690
691
692
693
694
695
696
697
698
699
700
701
702
703
704
705
706
707
708
709
710
711
712
713
714
715
716
717
718
719
720
721
722
723
724
725
726
727
728
729
730
731
732
733
734
735
736
737
738
739
740
741
742
743
744
745
746
747
748
749
750
751
752
753
754
755
756
757
758
759
760
761
762
763
764
765
766
767
768
769
770
771
772
773
774
775
776
777
778
779
780
781
782
783
784
785
786
787
788
789
790
791
792
793
794
795
796
797
798
799
800
801
802
803
804
805
806
807
808
809
810
811
812
813
814
815
816
817
818
819
820
821
822
823
824
825
826
827
828
829
830
831
832
833
834
835
836
837
838
839
840
841
842
843
844
845
846
847
848
849
850
851
852
853
854
855
856
857
858
859
860
861
862
863
864
865
866
867
868
869
870
871
872
873
874
875
876
877
878
879
880
881
882
883
884
885
886
887
888
889
890
891
892
893
894
895
896
897
898
899
900
901
902
903
904
905
906
907
908
909
910
911
912
913
914
915
916
917
918
919
920
921
922
923
924
925
926
927
928
929
930
931
932
933
934
935
936
937
938
939
940
941
942
943
944
945
946
947
948
949
950
951
952
953
954
955
956
957
958
959
960
961
962
963
964
965
966
967
968
969
970
971
972
973
974
975
976
977
978
979
980
981
982
983
984
985
986
987
988
989
990
991
992
993
994
995
996
997
998
999
1000
<!doctype html>
<html lang="en" class="no-js">
<head>
<meta charset="utf-8">
<meta name="viewport" content="width=device-width,initial-scale=1">
<meta name="author" content="I.T. Young & R. Ligteringen">
<link rel="shortcut icon" href="images/favicon.png">
<meta name="generator" content="mkdocs-1.1.2, mkdocs-material-6.0.2">
<title>8. Characterizing Signal-to-Noise Ratios - Introduction to Stochastic Signal Processing</title>
<link rel="stylesheet" href="assets/stylesheets/main.38780c08.min.css">
<link rel="stylesheet" href="assets/stylesheets/palette.3f72e892.min.css">
<link href="https://fonts.gstatic.com" rel="preconnect" crossorigin>
<link rel="stylesheet" href="https://fonts.googleapis.com/css?family=Georgia:300,400,400i,700%7CCourier&display=fallback">
<style>body,input{font-family:"Georgia",-apple-system,BlinkMacSystemFont,Helvetica,Arial,sans-serif}code,kbd,pre{font-family:"Courier",SFMono-Regular,Consolas,Menlo,monospace}</style>
<link rel="stylesheet" href="css/extra.css">
<link rel="stylesheet" href="css/imgtxt.css">
<link rel="stylesheet" href="css/tables.css">
<!-- Global site tag (gtag.js) - Google Analytics -->
<script async src="https://www.googletagmanager.com/gtag/js?id=G-8LE8Z88MMP"></script>
<script>
window.dataLayer = window.dataLayer || [];
function gtag(){dataLayer.push(arguments);}
gtag('js', new Date());
gtag('config', 'G-8LE8Z88MMP');
</script>
</head>
<body dir="ltr" data-md-color-scheme="" data-md-color-primary="none" data-md-color-accent="none">
<input class="md-toggle" data-md-toggle="drawer" type="checkbox" id="__drawer" autocomplete="off">
<input class="md-toggle" data-md-toggle="search" type="checkbox" id="__search" autocomplete="off">
<label class="md-overlay" for="__drawer"></label>
<div data-md-component="skip">
<a href="#characterizing-signal-to-noise-ratios" class="md-skip">
Skip to content
</a>
</div>
<div data-md-component="announce">
</div>
<!-- Application header -->
<header class="md-header" data-md-component="header">
<!-- Top-level navigation -->
<nav class="md-header-nav md-grid" aria-label="Header">
<!-- Link to home -->
<a
href=""
title="Introduction to Stochastic Signal Processing"
class="md-header-nav__button md-logo"
aria-label="Introduction to Stochastic Signal Processing"
>
<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 8a3 3 0 003-3 3 3 0 00-3-3 3 3 0 00-3 3 3 3 0 003 3m0 3.54C9.64 9.35 6.5 8 3 8v11c3.5 0 6.64 1.35 9 3.54 2.36-2.19 5.5-3.54 9-3.54V8c-3.5 0-6.64 1.35-9 3.54z"/></svg>
<!--
Insert an <img> here as an alternative for the standard logo if desired.
Then comment out the above two lines
<img src="images/favicon.png" style="margin-top:5px; border: 0px solid lime; border-radius:5px; width:120%; height: auto;" />
-->
</a>
<!-- Button to open drawer -->
<label class="md-header-nav__button md-icon" for="__drawer">
<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M3 6h18v2H3V6m0 5h18v2H3v-2m0 5h18v2H3v-2z"/></svg>
</label>
<!-- Header title -->
<div class="md-header-nav__title" data-md-component="header-title">
<a href="#jumpToBottom">
<span class="md-header-nav__topic">
Introduction to Stochastic Signal Processing
</span>
<span class="md-header-nav__topic">
8. Characterizing Signal-to-Noise Ratios
</span>
</a>
</div>
<!-- Button to open search dialogue -->
<label class="md-header-nav__button md-icon" for="__search">
<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M9.5 3A6.5 6.5 0 0116 9.5c0 1.61-.59 3.09-1.56 4.23l.27.27h.79l5 5-1.5 1.5-5-5v-.79l-.27-.27A6.516 6.516 0 019.5 16 6.5 6.5 0 013 9.5 6.5 6.5 0 019.5 3m0 2C7 5 5 7 5 9.5S7 14 9.5 14 14 12 14 9.5 12 5 9.5 5z"/></svg>
</label>
<!-- Search interface -->
<div class="md-search" data-md-component="search" role="dialog">
<label class="md-search__overlay" for="__search"></label>
<div class="md-search__inner" role="search">
<form class="md-search__form" name="search">
<input type="text" class="md-search__input" name="query" aria-label="Search" placeholder="Search" autocapitalize="off" autocorrect="off" autocomplete="off" spellcheck="false" data-md-component="search-query" data-md-state="active">
<label class="md-search__icon md-icon" for="__search">
<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M9.5 3A6.5 6.5 0 0116 9.5c0 1.61-.59 3.09-1.56 4.23l.27.27h.79l5 5-1.5 1.5-5-5v-.79l-.27-.27A6.516 6.516 0 019.5 16 6.5 6.5 0 013 9.5 6.5 6.5 0 019.5 3m0 2C7 5 5 7 5 9.5S7 14 9.5 14 14 12 14 9.5 12 5 9.5 5z"/></svg>
<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M20 11v2H8l5.5 5.5-1.42 1.42L4.16 12l7.92-7.92L13.5 5.5 8 11h12z"/></svg>
</label>
<button type="reset" class="md-search__icon md-icon" aria-label="Clear" data-md-component="search-reset" tabindex="-1">
<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M19 6.41L17.59 5 12 10.59 6.41 5 5 6.41 10.59 12 5 17.59 6.41 19 12 13.41 17.59 19 19 17.59 13.41 12 19 6.41z"/></svg>
</button>
</form>
<div class="md-search__output">
<div class="md-search__scrollwrap" data-md-scrollfix>
<div class="md-search-result" data-md-component="search-result">
<div class="md-search-result__meta">
Initializing search
</div>
<ol class="md-search-result__list"></ol>
</div>
</div>
</div>
</div>
</div>
<!-- Repository containing source -->
</nav>
</header>
<div class="md-container" data-md-component="container">
<main class="md-main" data-md-component="main">
<div class="md-main__inner md-grid">
<div class="md-sidebar md-sidebar--primary" data-md-component="navigation">
<div class="md-sidebar__scrollwrap">
<div class="md-sidebar__inner">
<nav class="md-nav md-nav--primary" aria-label="Navigation" data-md-level="0">
<label class="md-nav__title" for="__drawer">
<a href="" title="Introduction to Stochastic Signal Processing" class="md-nav__button md-logo" aria-label="Introduction to Stochastic Signal Processing">
<svg xmlns="http://www.w3.org/2000/svg" viewBox="0 0 24 24"><path d="M12 8a3 3 0 003-3 3 3 0 00-3-3 3 3 0 00-3 3 3 3 0 003 3m0 3.54C9.64 9.35 6.5 8 3 8v11c3.5 0 6.64 1.35 9 3.54 2.36-2.19 5.5-3.54 9-3.54V8c-3.5 0-6.64 1.35-9 3.54z"/></svg>
<!--
Insert an <img> here as an alternative for the standard logo if desired.
Then comment out the above two lines
<img src="images/favicon.png" style="margin-top:5px; border: 0px solid lime; border-radius:5px; width:120%; height: auto;" />
-->
</a>
Introduction to Stochastic Signal Processing
</label>
<ul class="md-nav__list" data-md-scrollfix>
<li class="md-nav__item">
<a href="Chap_1.html" class="md-nav__link">
1. How to use this iBook
</a>
</li>
<li class="md-nav__item">
<a href="Chap_2.html" class="md-nav__link">
2. Prologue
</a>
</li>
<li class="md-nav__item">
<a href="Chap_3.html" class="md-nav__link">
3. Introduction
</a>
</li>
<li class="md-nav__item">
<a href="Chap_4.html" class="md-nav__link">
4. Characterization of Random Signals
</a>
</li>
<li class="md-nav__item">
<a href="Chap_5.html" class="md-nav__link">
5. Correlations and Spectra
</a>
</li>
<li class="md-nav__item">
<a href="Chap_6.html" class="md-nav__link">
6. Filtering of Stochastic Signals
</a>
</li>
<li class="md-nav__item">
<a href="Chap_7.html" class="md-nav__link">
7. The Langevin Equation – A Case Study
</a>
</li>
<li class="md-nav__item md-nav__item--active">
<input class="md-nav__toggle md-toggle" data-md-toggle="toc" type="checkbox" id="__toc">
<label class="md-nav__link md-nav__link--active" for="__toc">
8. Characterizing Signal-to-Noise Ratios
<span class="md-nav__icon md-icon"></span>
</label>
<a href="Chap_8.html" class="md-nav__link md-nav__link--active">
8. Characterizing Signal-to-Noise Ratios
</a>
<nav class="md-nav md-nav--secondary" aria-label="Table of contents">
<label class="md-nav__title" for="__toc">
<span class="md-nav__icon md-icon"></span>
Table of contents
</label>
<ul class="md-nav__list" data-md-scrollfix>
<li class="md-nav__item">
<a href="#example-filtering-noise" class="md-nav__link">
Example: Filtering noise
</a>
</li>
<li class="md-nav__item">
<a href="#snr-for-deterministic-signals-in-the-presence-of-noise" class="md-nav__link">
SNR for deterministic signals in the presence of noise
</a>
</li>
<li class="md-nav__item">
<a href="#snr-for-random-signals-in-the-presence-of-noise" class="md-nav__link">
SNR for random signals in the presence of noise
</a>
<nav class="md-nav" aria-label="SNR for random signals in the presence of noise">
<ul class="md-nav__list">
<li class="md-nav__item">
<a href="#example-not-too-noisy" class="md-nav__link">
Example: Not too noisy
</a>
</li>
<li class="md-nav__item">
<a href="#example-cocktail-party-noise" class="md-nav__link">
Example: Cocktail party noise?
</a>
</li>
</ul>
</nav>
</li>
<li class="md-nav__item">
<a href="#snr-for-signals-and-systems-with-poisson-noise" class="md-nav__link">
SNR for signals and systems with Poisson noise
</a>
</li>
<li class="md-nav__item">
<a href="#problems" class="md-nav__link">
Problems
</a>
<nav class="md-nav" aria-label="Problems">
<ul class="md-nav__list">
<li class="md-nav__item">
<a href="#problem-81" class="md-nav__link">
Problem 8.1
</a>
</li>
<li class="md-nav__item">
<a href="#problem-82" class="md-nav__link">
Problem 8.2
</a>
</li>
<li class="md-nav__item">
<a href="#problem-83" class="md-nav__link">
Problem 8.3
</a>
</li>
</ul>
</nav>
</li>
</ul>
</nav>
</li>
<li class="md-nav__item">
<a href="Chap_9.html" class="md-nav__link">
9. The Matched Filter
</a>
</li>
<li class="md-nav__item">
<a href="Chap_10.html" class="md-nav__link">
10. The Wiener filter
</a>
</li>
<li class="md-nav__item">
<a href="Chap_11.html" class="md-nav__link">
11. Aspects of Estimation
</a>
</li>
<li class="md-nav__item">
<a href="Chap_12.html" class="md-nav__link">
12. Spectral Estimation
</a>
</li>
<li class="md-nav__item">
<a href="Chap_13.html" class="md-nav__link">
Appendices
</a>
</li>
<li class="md-nav__item">
<a href="info.html" class="md-nav__link">
Information
</a>
</li>
</ul>
</nav>
</div>
</div>
</div>
<div class="md-sidebar md-sidebar--secondary" data-md-component="toc">
<div class="md-sidebar__scrollwrap">
<div class="md-sidebar__inner">
<nav class="md-nav md-nav--secondary" aria-label="Table of contents">
<label class="md-nav__title" for="__toc">
<span class="md-nav__icon md-icon"></span>
Table of contents
</label>
<ul class="md-nav__list" data-md-scrollfix>
<li class="md-nav__item">
<a href="#example-filtering-noise" class="md-nav__link">
Example: Filtering noise
</a>
</li>
<li class="md-nav__item">
<a href="#snr-for-deterministic-signals-in-the-presence-of-noise" class="md-nav__link">
SNR for deterministic signals in the presence of noise
</a>
</li>
<li class="md-nav__item">
<a href="#snr-for-random-signals-in-the-presence-of-noise" class="md-nav__link">
SNR for random signals in the presence of noise
</a>
<nav class="md-nav" aria-label="SNR for random signals in the presence of noise">
<ul class="md-nav__list">
<li class="md-nav__item">
<a href="#example-not-too-noisy" class="md-nav__link">
Example: Not too noisy
</a>
</li>
<li class="md-nav__item">
<a href="#example-cocktail-party-noise" class="md-nav__link">
Example: Cocktail party noise?
</a>
</li>
</ul>
</nav>
</li>
<li class="md-nav__item">
<a href="#snr-for-signals-and-systems-with-poisson-noise" class="md-nav__link">
SNR for signals and systems with Poisson noise
</a>
</li>
<li class="md-nav__item">
<a href="#problems" class="md-nav__link">
Problems
</a>
<nav class="md-nav" aria-label="Problems">
<ul class="md-nav__list">
<li class="md-nav__item">
<a href="#problem-81" class="md-nav__link">
Problem 8.1
</a>
</li>
<li class="md-nav__item">
<a href="#problem-82" class="md-nav__link">
Problem 8.2
</a>
</li>
<li class="md-nav__item">
<a href="#problem-83" class="md-nav__link">
Problem 8.3
</a>
</li>
</ul>
</nav>
</li>
</ul>
</nav>
</div>
</div>
</div>
<div class="md-content">
<article class="md-content__inner md-typeset">
<h1 id="characterizing-signal-to-noise-ratios">Characterizing Signal-to-Noise Ratios<a class="headerlink" href="#characterizing-signal-to-noise-ratios" title="Permanent link">¶</a></h1>
<p>We now know how to characterize stochastic signals and how to determine
how that characterization changes as a stochastic signal is passed
through an LTI system, a linear filter. We start with an example.</p>
<h4 id="example-filtering-noise">Example: Filtering noise<a class="headerlink" href="#example-filtering-noise" title="Permanent link">¶</a></h4>
<p>Let the input to an LTI system be given by an ergodic process <span class="arithmatex">\(x[n]\)</span> with power spectrum <span class="arithmatex">\({S_{xx}}(\Omega ).\)</span> Then <span class="arithmatex">\({S_{yy}}(\Omega ) = {\left| {H( - \Omega )} \right|^2}{S_{xx}}(\Omega ).\)</span> If <span class="arithmatex">\({\varphi _{xx}}[k] = {N_o}\delta [k]\)</span> (white noise input) and <span class="arithmatex">\(h[n] = {\alpha ^n}u[n]\)</span> with <span class="arithmatex">\(\alpha\)</span> <em>real</em> and <span class="arithmatex">\(\left| \alpha \right| < 1\)</span> then:</p>
<div class="" id="eq:snreq1">
<table class="eqTable">
<tr>
<td class="eqTableTag">(8.1)</td>
<td class="eqTableEq">
<div>$$\begin{array}{l}
H(\Omega ) = \frac{1}{{1 - \alpha {e^{ - j\Omega }}}}\,\,\,\,\,\,\,\,\,\,\,{S_{xx}}(\Omega ) = {N_0}\,\,\\
{S_{yy}}(\Omega ) = {N_0}{\left| {\frac{1}{{1 - \alpha {e^{ - j\Omega }}}}} \right|^2} = \frac{{{N_0}}}{{1 + {\alpha ^2} - 2\alpha \cos \Omega }}
\end{array}$$</div>
</td>
</tr>
</table>
</div>
<p>The output random process <span class="arithmatex">\(y[n]\)</span> is a non-white, random process.</p>
<p>In the following sections we will look at several applications of these ideas as well as attempt to understand why the power density spectrum and the correlation functions play such an important role in stochastic signal processing. Further, we will focus our attention on those situations where the model is additive noise that is statistically
independent of the signal.</p>
<p>We begin by defining a well-known measure of system performance, the signal-to-noise ratio (<i>SNR</i>). We make a distinction between two cases 1) where the signal is deterministic or 2) where the signal is stochastic. In either case, of course, the noise is considered to be stochastic.</p>
<h2 id="snr-for-deterministic-signals-in-the-presence-of-noise">SNR for deterministic signals in the presence of noise<a class="headerlink" href="#snr-for-deterministic-signals-in-the-presence-of-noise" title="Permanent link">¶</a></h2>
<p>For deterministic signals, signals where the signal amplitude is known for every time instant, we define the signal-to noise ratio as:</p>
<div class="" id="eq:snreq2">
<table class="eqTable">
<tr>
<td class="eqTableTag">(8.2)</td>
<td class="eqTableEq">
<div>$$S\tilde N{R_d} = \frac{S}{N} = \frac{{\max \left| {y[n]} \right|}}{{{\sigma _{nn}}}}$$</div>
</td>
</tr>
</table>
</div>
<p>or using the logarithmic (Bode) description:</p>
<div class="" id="eq:snreq3">
<table class="eqTable">
<tr>
<td class="eqTableTag">(8.3)</td>
<td class="eqTableEq">
<div>$$SN{R_d} = 20\,{\log _{10}}\left( {\frac{{\max \left| {y[n]} \right|}}{{{\sigma _{nn}}}}} \right)$$</div>
</td>
</tr>
</table>
</div>
<h2 id="snr-for-random-signals-in-the-presence-of-noise">SNR for random signals in the presence of noise<a class="headerlink" href="#snr-for-random-signals-in-the-presence-of-noise" title="Permanent link">¶</a></h2>
<p>For random signals (such as speech) which bear information but whose amplitudes cannot be predicted with certainty, we define the signal-to-noise ratio as:</p>
<div class="" id="eq:snreq4">
<table class="eqTable">
<tr>
<td class="eqTableTag">(8.4)</td>
<td class="eqTableEq">
<div>$$S\tilde N{R_r} = \frac{S}{N} = \frac{{{\sigma _{ss}}}}{{{\sigma _{nn}}}}$$</div>
</td>
</tr>
</table>
</div>
<div class="" id="eq:snreq5">
<table class="eqTable">
<tr>
<td class="eqTableTag">(8.5)</td>
<td class="eqTableEq">
<div>$$SN{R_r} = 20\,{\log _{10}}\left( {\frac{{{\sigma _{ss}}}}{{{\sigma _{nn}}}}} \right)$$</div>
</td>
</tr>
</table>
</div>
<p>In both cases <span class="arithmatex">\({\sigma _{nn}}\)</span> is the standard deviation of the noise. In the case of
random, information-bearing signals, <span class="arithmatex">\({\sigma _{ss}}\)</span> is the standard deviation of the signal amplitudes.</p>
<p>From our earlier result, <a href="Chap_5.html#eq:covariances">Equation 5.2</a>, we know that</p>
<div class="" id="eq:snreq6">
<table class="eqTable">
<tr>
<td class="eqTableTag">(8.6)</td>
<td class="eqTableEq">
<div>$$\sigma _{nn}^2 = {\gamma _{nn}}[0] = {\varphi _{nn}}[0] - {\left| {{m_n}} \right|^2}$$</div>
</td>
</tr>
</table>
</div>
<p>We begin with the case where <span class="arithmatex">\({m_n} = 0\)</span> giving <span class="arithmatex">\({\sigma _{nn}} = \sqrt {{\varphi _{nn}}[0]}.\)</span> Noise with zero mean leads to a formulation for the standard deviation of the
noise in terms of the autocorrelation function of the noise evaluated at <span class="arithmatex">\(k = 0.\)</span> To simplify matters with respect to the case of information-bearing random signals, we will assume that their mean is also zero yielding <span class="arithmatex">\({\sigma _{ss}} = \sqrt {{\varphi _{ss}}[0]}.\)</span></p>
<p>The justification for this is simple. If the mean is not zero then we can express the random signal <span class="arithmatex">\(y[n]\)</span> as <span class="arithmatex">\(y[n] = {m_y} + {y_0}[n]\)</span> where <span class="arithmatex">\({m_y}\)</span> is the (deterministic) non-zero average and <span class="arithmatex">\({y_0}[n]\)</span> is a stochastic signal with the same statistics as <span class="arithmatex">\(y[n]\)</span> except that its average is zero. For our discussion of <i>SNR</i>, we, therefore, focus on the
zero-mean stochastic term.</p>
<p>We can now rewrite our definition using our zero-mean assumption as:</p>
<div class="mainresult" id="eq:snreq7">
<table class="eqTable">
<tr>
<td class="eqTableTag">(8.7)</td>
<td class="eqTableEq">
<div>$$SN{R_d} = 10\,{\log _{10}}\left( {\frac{{\max {{\left| {y[n]} \right|}^2}}}{{{\varphi _{nn}}[0]}}} \right)$$</div>
</td>
</tr>
</table>
</div>
<div class="mainresult" id="eq:snreq8">
<table class="eqTable">
<tr>
<td class="eqTableTag">(8.8)</td>
<td class="eqTableEq">
<div>$$SN{R_r} = 10\,{\log _{10}}\left( {\frac{{{\varphi _{ss}}[0]}}{{{\varphi _{nn}}[0]}}} \right)$$</div>
</td>
</tr>
</table>
</div>
<h4 id="example-not-too-noisy">Example: Not too noisy<a class="headerlink" href="#example-not-too-noisy" title="Permanent link">¶</a></h4>
<p>To illustrate the concept for a deterministic signal, say that a video signal generated by a known test pattern is confined to the interval 0 to 1 volt and that the noise, <span class="arithmatex">\(N[n],\)</span> contaminating the video signal is 10 mv RMS. <i>RMS</i> means “root-mean-square” or <span class="arithmatex">\(\sqrt {E\left\{ {{{\left| {N[n]} \right|}^2}} \right\}}.\)</span> From our zero mean assumption, 10 mv RMS is equivalent to <span class="arithmatex">\({\sigma _{nn}} = 10\)</span> mv. According to our definition:</p>
<div class="" id="eq:snreq9">
<table class="eqTable">
<tr>
<td class="eqTableTag">(8.9)</td>
<td class="eqTableEq">
<div>$$\begin{array}{*{20}{l}}
{SN{R_d}}&{ = 10\,{{\log }_{10}}\left( {\frac{{{{\left( {1\,{\rm{V}}} \right)}^2}}}{{{{\left( {{{10}^{ - 2}}\;{\rm{V}}} \right)}^2}}}} \right)}\\
{\,\,\,}&{ = 10\,{{\log }_{10}}\left( {{{10}^4}} \right) = 40\;{\rm{dB}}}
\end{array}$$</div>
</td>
</tr>
</table>
</div>
<h4 id="example-cocktail-party-noise">Example: Cocktail party noise?<a class="headerlink" href="#example-cocktail-party-noise" title="Permanent link">¶</a></h4>
<p>We can also illustrate the concept for a random signal as follows. Let the discrete-time output of a microphone be described by <span class="arithmatex">\(r[n] = s[n] + N[n]\)</span> where <span class="arithmatex">\(s[n]\)</span> is a sampled, real speech signal and <span class="arithmatex">\(N[n]\)</span> is sampled, real random noise from the electronics. This is
shown in <a href="#fig:stand_model1">Figure 8.1</a>.</p>
<figure class="figaltcap fullsize" id="fig:stand_model1"><img src="images/Fig_8_1.gif" /><figcaption><strong>Figure 8.1:</strong> Model for a random signal speech, <span class="arithmatex">\(s[n\rbrack,\)</span> contaminated by additive noise <span class="arithmatex">\(N[n\rbrack\)</span> and resulting in <span class="arithmatex">\(r[n\rbrack.\)</span></figcaption>
</figure>
<p>The random speech signal <span class="arithmatex">\(s[n]\)</span> has power density spectrum <span class="arithmatex">\({S_{ss}}(\Omega )\)</span> given by:</p>
<div class="" id="eq:snreq10">
<table class="eqTable">
<tr>
<td class="eqTableTag">(8.10)</td>
<td class="eqTableEq">
<div>$${S_{ss}}(\Omega ) = \frac{{{S_0}}}{{1 + {\alpha ^2} - 2\alpha \cos \Omega }}$$</div>
</td>
</tr>
</table>
</div>
<p>and the independent, additive noise <span class="arithmatex">\(N[n]\)</span> is characterized by:</p>
<div class="" id="eq:snreq11">
<table class="eqTable">
<tr>
<td class="eqTableTag">(8.11)</td>
<td class="eqTableEq">
<div>$${S_{nn}}(\Omega ) = \frac{{{N_0}}}{{1 + {\beta ^2} - 2\beta \cos \Omega }}$$</div>
</td>
</tr>
</table>
</div>
<p>The <i>SÑR<sub>r</sub></i> will then follow from:</p>
<div class="" id="eq:snreq12">
<table class="eqTable">
<tr>
<td class="eqTableTag">(8.12)</td>
<td class="eqTableEq">
<div>$$\begin{array}{*{20}{l}}
{{\varphi _{ss}}[0]}&{ = \frac{1}{{2\pi }}\int\limits_{ - \pi }^{ + \pi } {{S_{ss}}(\Omega )} d\Omega }\\
{\,\,\,}&{ = \frac{1}{{2\pi }}\int\limits_{ - \pi }^{ + \pi } {\frac{{{S_0}}}{{1 + {\alpha ^2} - 2\alpha \cos \Omega }}} d\Omega }\\
{\,\,\,}&{ = \frac{{{S_0}}}{{1 - {\alpha ^2}}}}
\end{array}$$</div>
</td>
</tr>
</table>
</div>
<div class="" id="eq:snreq13">
<table class="eqTable">
<tr>
<td class="eqTableTag">(8.13)</td>
<td class="eqTableEq">
<div>$$\begin{array}{*{20}{l}}
{{\varphi _{nn}}[0]}&{ = \frac{1}{{2\pi }}\int\limits_{ - \pi }^{ + \pi } {\frac{{{N_0}}}{{1 + {\beta ^2} - 2\beta \cos \Omega }}} d\Omega }\\
{\,\,\,}&{ = \frac{{{N_0}}}{{1 - {\beta ^2}}}}
\end{array}$$</div>
</td>
</tr>
</table>
</div>
<p>The proof of <a href="Chap_8.html#eq:snreq12">Equation 8.12</a> can be found in <a href="Chap_8.html#problem-81">Problem 8.1</a>. Using <a href="Chap_8.html#eq:snreq8">Equation 8.8</a>:</p>
<div class="" id="eq:snreq14">
<table class="eqTable">
<tr>
<td class="eqTableTag">(8.14)</td>
<td class="eqTableEq">
<div>$$\begin{array}{*{20}{l}}
{SN{R_r}}&{ = 10\,{{\log }_{10}}\left( {\frac{{{S_o}/\left( {1 - {\alpha ^2}} \right)}}{{{N_o}/\left( {1 - {\beta ^2}} \right)}}} \right)}\\
{\,\,\,}&{ = 10\,{{\log }_{10}}\left( {\frac{{{S_0}}}{{{N_0}}}\,\left( {\frac{{1 - {\beta ^2}}}{{1 - {\alpha ^2}}}} \right)} \right)}
\end{array}$$</div>
</td>
</tr>
</table>
</div>
<p>The two power density spectra are illustrated in <a href="#fig:stand_model2">Figure 8.2</a>.</p>
<figure class="figaltcap fullsize" id="fig:stand_model2"><img src="images/Fig_8_2.png" /><figcaption><strong>Figure 8.2:</strong> Power density spectra of <strong>signal</strong> <span class="arithmatex">\({S_{ss}}(\Omega )\)</span> (in <strong><font color="red">red</font></strong>) and <strong>noise</strong> <span class="arithmatex">\({S_{nn}}(\Omega )\)</span> (in <strong><font color="#00bb00">green</font></strong>) for the values <span class="arithmatex">\(\alpha = 4/7\)</span> and <span class="arithmatex">\(\beta = - 2/3.\)</span></figcaption>
</figure>
<p>For <span class="arithmatex">\({S_0} = 1,\)</span> <span class="arithmatex">\({N_0} = 1,\)</span> <span class="arithmatex">\(\alpha = 4/7,\)</span> and <span class="arithmatex">\(\beta = - 2/3,\)</span> the <i>SNR<sub>r</sub></i> = –0.84 dB.</p>
<h2 id="snr-for-signals-and-systems-with-poisson-noise">SNR for signals and systems with Poisson noise<a class="headerlink" href="#snr-for-signals-and-systems-with-poisson-noise" title="Permanent link">¶</a></h2>
<p>One very important class of noise is Poisson noise. In this case, the noise is not Gaussian, not additive, not independent of the signal, and has a mean value that is greater than zero. We would be tempted to ignore it as a mathematical oddity were it not a type of noise that is encountered in a variety of physical systems.</p>
<p>A <a href="https://en.wikipedia.org/wiki/Poisson_process">Poisson process</a> is one that describes the probability of <span class="arithmatex">\(n\)</span> independent discrete events occurring in a fixed amount of time <span class="arithmatex">\(T\)</span> if the time between occurrences of successive events is described by an exponential-type probability density function with a rate-parameter <span class="arithmatex">\(\lambda.\)</span> The units of <span class="arithmatex">\(\lambda\)</span> are events/(unit time). The value of <span class="arithmatex">\(T\)</span>—the length of the observation window or the integration time—is frequently known. An exceptionally lucid derivation of the Poisson distribution can be found in Section 18.13 of Thomas<sup id="fnref:thomas1960"><a class="footnote-ref" href="#fn:thomas1960">1</a></sup>.</p>
<p>The emission of fluorescence photons in a labeled biological specimen, the production of photoelectrons in a CCD camera due to thermal vibrations (dark noise or dark current), and the radioactive decay of a population of identical atoms are examples of random processes governed by a Poisson distribution.</p>
<p>The probability of <span class="arithmatex">\(n\)</span> events in <span class="arithmatex">\(T\)</span> units of time is given by:</p>
<div class="" id="eq:snreq15">
<table class="eqTable">
<tr>
<td class="eqTableTag">(8.15)</td>
<td class="eqTableEq">
<div>$$p(n|\lambda T) = \frac{{{{\left( {\lambda T} \right)}^n}{e^{ - \lambda T}}}}{{n!}}$$</div>
</td>
</tr>
</table>
</div>
<p>For Poisson processes the signal is intrinsically noisy and needs no “help” from external sources. We define the signal-to-noise ratio as:</p>
<div class="" id="eq:snreq16">
<table class="eqTable">
<tr>
<td class="eqTableTag">(8.16)</td>
<td class="eqTableEq">
<div>$$\begin{array}{*{20}{l}}
{S\tilde N{R_P} = \frac{S}{N} = \frac{\mu }{\sigma }}\\
{SN{R_P} = 20\,{{\log }_{10}}\left( {\frac{\mu }{\sigma }} \right)}
\end{array}$$</div>
</td>
</tr>
</table>
</div>
<p>where <span class="arithmatex">\(\mu\)</span> is the expected value of the stochastic signal and <span class="arithmatex">\(\sigma\)</span> is its
standard deviation. This process—whether one is talking about <span class="arithmatex">\(\lambda\)</span> or <span class="arithmatex">\(\lambda T\)</span>—is
a one-parameter process. This means that both <span class="arithmatex">\(\mu\)</span> and <span class="arithmatex">\(\sigma\)</span> will be defined by
<span class="arithmatex">\(\lambda\)</span> (or <span class="arithmatex">\(\lambda T\)</span>).</p>
<p>From <a href="Chap_3.html#problem-34">Problem 3.4c</a> we know that <span class="arithmatex">\(\mu = \lambda T\)</span> and
<span class="arithmatex">\(\sigma = \sqrt {\lambda T}.\)</span> This means that:</p>
<div class="" id="eq:snreq17">
<table class="eqTable">
<tr>
<td class="eqTableTag">(8.17)</td>
<td class="eqTableEq">
<div>$$\begin{array}{*{20}{l}}
{S\tilde N{R_P} = \frac{{\lambda T}}{{\sqrt {\lambda T} }} = \sqrt {\lambda T} }\\
{SN{R_P} = 20{{\log }_{10}}\left( {\sqrt {\lambda T} } \right) = 10\,{{\log }_{10}}\left( {\lambda T} \right)}
\end{array}$$</div>
</td>
</tr>
</table>
</div>
<p>If we wish to increase the signal-to-noise ratio, we either have to increase the rate parameter <span class="arithmatex">\(\lambda\)</span> and/or use a longer observation window <span class="arithmatex">\(T.\)</span> Either approach simply means that we need more data.</p>
<p>The implications of these results warrant some discussion. As the strength of a Poisson signal increases—for example by using a source that emits a greater number of photons per second—so does the strength of the fluctuations. The former is measured by <span class="arithmatex">\(\mu\)</span> and the latter by <span class="arithmatex">\(\sigma.\)</span> This is illustrated in <a href="#fig:stand_model3">Figure 8.3</a>.</p>
<figure class="figaltcap fullsize" id="fig:stand_model3"><img src="images/Fig_8_3.gif" /><figcaption><strong>Figure 8.3:</strong> (left) The top half of the image is a “clean” image where the intensity in each block increases by 32 brightness levels. The bottom half is Poisson noise where the mean value (<span class="arithmatex">\(\mu\)</span>) in each block is the same as the corresponding block above it. The <strong><font color="red">red</font></strong> line is one row of the clean image and the <strong><font color="blue">blue</font></strong> line is one row of the noisy image. (right) The intensities along the <strong><font color="red">red</font></strong> line and <strong><font color="blue">blue</font></strong> line versus the column numbers (with column <span class="arithmatex">\(0\)</span> being in the middle) are displayed in <strong><font color="red">red</font></strong> and <strong><font color="blue">blue</font></strong>, respectively.</figcaption>
</figure>
<p>In the left-hand panel you should observe that the noise increases even as the average intensity increases. The basis for this observation will be discussed in <a href="Chap_8.html#problem-82">Problem 8.2</a>. In the right-hand panel we see the fluctuations increase as the average increases.
The calculation in <a href="Chap_8.html#eq:snreq17">Equation 8.17</a> shows, however, that as the source strength increases (from left to right) the <i>SNR</i> increases as well.</p>
<p>There is much more that can be said about Poisson noise; we will address the issue of the estimation of <span class="arithmatex">\(\lambda\)</span> (or <span class="arithmatex">\(\lambda T\)</span>) in an example in <a href="Chap_11.html#example-estimating-the-poisson-rate">Chapter 11</a>. </p>
<h2 class="problems" id="problems">Problems<a class="headerlink" href="#problems" title="Permanent link">¶</a></h2>
<h3 id="problem-81">Problem 8.1<a class="headerlink" href="#problem-81" title="Permanent link">¶</a></h3>
<p>The autocorrelation function of a random process is given by:</p>
<div class="" id="eq:snreq18">
<table class="eqTable">
<tr>
<td class="eqTableTag">(8.18)</td>
<td class="eqTableEq">
<div>$${\varphi _{xx}}[k] = {a^{\left| k \right|}}\,\,\,\,\,\,\,\,\,\,\left| a \right| < 1$$</div>
</td>
</tr>
</table>
</div>
<ol>
<li>Determine the power spectral density <span class="arithmatex">\({S_{xx}}(\Omega ).\)</span></li>
<li>Using <span class="arithmatex">\({S_{xx}}(\Omega )\)</span> and Fourier properties prove <a href="Chap_8.html#eq:snreq12">Equation 8.12</a>.</li>
<li>What is <span class="arithmatex">\({m_x}\)</span> for this process?<a class="indentlist" href=""></a> </li>
</ol>
<h3 id="problem-82">Problem 8.2<a class="headerlink" href="#problem-82" title="Permanent link">¶</a></h3>
<p>It has been recognized for over 150 years that the perception <span class="arithmatex">\(p\)</span> of a physical stimulus <span class="arithmatex">\(S\)</span> is described in many cases by:</p>
<div class="" id="eq:snreq19">
<table class="eqTable">
<tr>
<td class="eqTableTag">(8.19)</td>
<td class="eqTableEq">
<div>$$\Delta p = k\frac{{\Delta S}}{S}$$</div>
</td>
</tr>
</table>
</div>
<p>where <span class="arithmatex">\(\Delta p\)</span> is the change in perception, <span class="arithmatex">\(k\)</span> is a proportionality constant, and <span class="arithmatex">\(\Delta S\)</span> represents a change in the stimulus. This is know as the <a href="https://en.wikipedia.org/wiki/Weber%E2%80%93Fechner_law">Weber-Fechner Law</a>.</p>
<ol>
<li>As we proceed to smaller increments in the stimulus, that is as <span class="arithmatex">\(\Delta S \to 0,\)</span> what is the functional relation between <span class="arithmatex">\(p\)</span> and <span class="arithmatex">\(S,\)</span> that is, what is <span class="arithmatex">\(p(S)?\)</span></li>
<li>For the stimulus shown in <a href="#fig:stand_model3">Figure 8.3</a>, how does our observation of the noise relate to <i>SNR<sub>p</sub></i>?<a class="indentlist" href=""></a> </li>
</ol>
<p>More information about the application of the Weber-Fechner Law to visual perception can be found in Hecht<sup id="fnref:hecht1924"><a class="footnote-ref" href="#fn:hecht1924">2</a></sup>.</p>
<h3 id="problem-83">Problem 8.3<a class="headerlink" href="#problem-83" title="Permanent link">¶</a></h3>
<p>The light, that passes through a test tube containing a suspension of particles in liquid, is measured by a photomultiplier and converted to a discrete-time (digital) signal. The output signal <span class="arithmatex">\(s[n]\)</span> from the photomultiplier shows a stochastic variation due to the sum of two statistically independent sources: 1) the Brownian motion <span class="arithmatex">\(B[n]\)</span> of the particles in the liquid, and, 2) the noise from the photomultiplier <span class="arithmatex">\(N[n].\)</span> Thus:</p>
<div class="" id="eq:snreq20">
<table class="eqTable">
<tr>
<td class="eqTableTag">(8.20)</td>
<td class="eqTableEq">
<div>$$s[n] = B[n] + N[n]$$</div>
</td>
</tr>
</table>
</div>
<p>The autocorrelation function of the Brownian motion, that is the velocity, is modeled by:</p>
<div class="" id="eq:snreq21">
<table class="eqTable">
<tr>
<td class="eqTableTag">(8.21)</td>
<td class="eqTableEq">
<div>$${\varphi _{BB}}[k] = {e^{ - \left| k \right|/D}} = {\left( {{e^{ - 1/D}}} \right)^{\left| k \right|}} = {\varepsilon ^{\left| k \right|}}$$</div>
</td>
</tr>
</table>
</div>
<p>with <span class="arithmatex">\(0 < \varepsilon < 1.\)</span> The autocorrelation function of the photomultiplier noise is modeled by:</p>
<div class="" id="eq:snreq22">
<table class="eqTable">
<tr>
<td class="eqTableTag">(8.22)</td>
<td class="eqTableEq">
<div>$${\varphi _{NN}}[k] = \delta [k]$$</div>
</td>
</tr>
</table>
</div>
<ol>
<li>What is the autocorrelation function of the signal <span class="arithmatex">\(s[n]\)</span> from the photomultiplier, that is, <span class="arithmatex">\({\varphi _{ss}}[k]?\)</span></li>
<li>What is the power spectral density <span class="arithmatex">\({S_{ss}}(\Omega )\)</span> of the measured signal? Sketch <span class="arithmatex">\({S_{ss}}(\Omega ).\)</span><a class="indentlist" href=""></a> </li>
</ol>
<p>We wish to estimate <span class="arithmatex">\(D,\)</span> a parameter related to the diffusion of the particles in the liquid. Our estimate of <span class="arithmatex">\(D\)</span> will, of course, be sensitive to the signal-to-noise ratio <i>SÑR<sub>r</sub></i> as defined in <a href="Chap_8.html#eq:snreq4">Equation 8.4</a>.</p>
<ol start="3">
<li>Determine the value of <i>SÑR<sub>r</sub></i>.<a class="indentlist" href=""></a> </li>
</ol>
<p>To improve the signal-to-noise ratio we first filter the signal <span class="arithmatex">\(s[n]\)</span> as shown below:</p>
<figure class="figaltcap fullsize" id="fig:stand_model4"><img src="images/Fig_P8_3.gif" /><figcaption><strong>Figure 8.4:</strong> Standard model of an (LTI) filter applied <em>after</em> a signal has been contaminated by additive noise.</figcaption>
</figure>
<p>where <span class="arithmatex">\(h[n]\)</span> is an ideal low-pass filter whose Fourier transform is
specified in the baseband <span class="arithmatex">\(- \pi \lt \Omega \le + \pi\)</span> as:</p>
<div class="" id="eq:snreq23">
<table class="eqTable">
<tr>
<td class="eqTableTag">(8.23)</td>
<td class="eqTableEq">
<div>$$H(\Omega ) = {\mathscr{F}}\left\{ {h[n]} \right\} = \left\{ {\begin{array}{*{20}{c}}
1&{\left| \Omega \right| \le \pi /2}\\
0&{\left| \Omega \right| > \pi /2}
\end{array}} \right.$$</div>
</td>
</tr>
</table>
</div>
<ol start="4">
<li>Determine <i>SÑR<sub>q</sub></i> from the filtered version of the signal, <span class="arithmatex">\(q[n].\)</span> Under what conditions will the signal-to-noise ratio improve, that is, increase as we go from <span class="arithmatex">\(s[n]\)</span> to <span class="arithmatex">\(q[n]?\)</span><a class="indentlist" href=""></a> </li>
</ol>
<!--
## Laboratory Exercises {: .labexp }
### Laboratory Exercise 8.1
<table class="imgtxt">
<tr>
<td>
<div><a href='LabExps/Lab_8.1.html'>
<img src='images/SNR_8.1.gif' width=auto height=auto
style="padding:2px; border:2px solid steelblue; border-radius:11px;"></a>
</div>
</td>
<td>
The camera in your device is not perfect. But just how good (or bad) is it?
In this laboratory exercise you will measure the noise characteristics of
your camera system. To start the exercise, click on the icon to the left.
</td>
</tr>
</table>
### Laboratory Exercise 8.2
<table class="imgtxt">
<tr>
<td>
<div><a href='LabExps/Lab_8.2.html'>
<img src='images/SNR_8.2.gif' width=auto height=auto
style="padding:2px; border:2px solid steelblue; border-radius:11px;"></a>
</div>
</td>
<td>
We can use the tools that we have already seen to look at the amplitude distribution of a difference image, its power spectral density, and its autocorrelation. To start the exercise, click on the icon to the left.
</td>
</tr>
</table>
### Laboratory Exercise 8.3
<table class="imgtxt">
<tr>
<td>
<div><a href='LabExps/Lab_8.3.html'>
<img src='images/SNR_8.3.gif' width=auto height=auto
style="padding:2px; border:2px solid steelblue; border-radius:11px;"></a>
</div>
</td>
<td>
We examine the noise characteristics of the microphone system of your device
with the help of the average volume level, the autocorrelation function, and
the power spectral density when “no” acoustic signal is proffered. To start
the exercise, click on the icon to the left.
</td>