166

|

5 Methods for Discrete Processing and Analysis of Biosignals

0

1

2

3

4

5

time t /s

-1

-0.5

0

0.5

1

amplitude

without noise

with noise

0

2

4

6

8

10

frequency f / Hz

-300

-200

-100

0

amplitude / dB

without noise

with noise

0

1

2

3

4

5

time t /s

-1

-0.5

0

0.5

1

amplitude

averaged

with noise

0

2

4

6

8

10

frequency f / Hz

-300

-200

-100

0

amplitude / dB

averaged

with noise

0

1

2

3

4

5

time t /s

-1

-0.5

0

0.5

1

amplitude

averaged

with noise

0

2

4

6

8

10

frequency f / Hz

-300

-200

-100

0

amplitude / dB

averaged

with noise

0

5

10

15

20

25

30

35

40

45

50

Number of averages N

5

10

15

20

25

SNR / dBc

Fig. 5.12: Coherent averaging using the example of a noisy periodic signal (top left): after a 5-fold

(centre left) and after a 50-fold coherent averaging (bottom left); the spectra of the signals compared

to the non-noisy signal (top right) and the noisy signals (centre right, bottom right) clearly shows

the reduction of the signal-noise-distance. The greater the number of averages N, the better the SNR

in dBc (bottom).

%% graphical representation of the SNR

subplot(4,2,7:8);

plot(1:N,SNR)

xlabel('Number of averages N');

ylabel('SNR / dBc');