er function
ously.
7.3. Some Examples Using Correlations Estimated by Time Averaging
Autocorrelations are calculated routinely in applications involving radar, geo-
physics, sonar, adaptive systems, communications, and the like. The similarity between
two waveforms may be measured using correlation methods by multiplying the two
waveforms, ordinate by ordinate, and then adding the products over the duration of the
waveforms, as we did in the last example, The similarity between the two waveforms is
assessed by the size of the sum of the products. Ifthe waveforms are identical (or nearly
so), then nearly every product contributes a positive term to the sum. The sum will
therefore be large. If, however, the waveforms are dissimilar, then some positive prod-
ucts may be canceled by negative products in the sum, and the final result will be small.
‘Ariother application of correlation i the measurement ofthe time delay ofa signal, es we consi
ered in Example 6.19. Imagine we have two identical waveforms, but one is shifted in time rela
tive tothe other, asin Figure 7.5, except that the shift isa delay, tht is, shifted to the right rather
than tothe left. If we calculate the crosscorrelation function for these two waveforms, we wil find
‘thatthe erosscorelation function will peak atthe time sift corresponding tothe time delay be-
tween the two waveforms, We may measure the instant at which the peak occurs in the crosscor-
relation function to determine the time delay.
sbi. mip, sum, and vide by N oobi Ryy(2)
Rea
No
Figure7.4 —_Exomple of estimating an aviocoreation function
from dato
| Example 7.9