2. SPECTRA OF CONTINUOUS AND DISCRETE SIGNALS
The spectral density function (spectrum) of a given
continuous or discrete process is defined as the Fourier
transform of the autocorrelation function of this process.
The spectrum is a function that measures the average rates
of fluctuation of a given set of signals. It represents,
therefore , the distribution of power with respect to
frequency . The spectrum of a continuous process can be
determined as follows :
P(e) = j| ga) ej?* ak
z 5 R,(k) { cos wk - j sin wk Ï dk (1)
Where :
Fat?) is the spectral density at frequency co
R (X) is the autocorrelation function at lag k.
As the autocorrelation function is always an even function,
the product R (k) . sin «w k integrates to zero , therefore,
PL.) is alwäys real(Schwartz,et 81,1975)
Fl) a em Re Ck) cos Wk dk (2)
For O < w ££ © , the inverse Fourier transform will give :
à
Ry) = = ) mq) coesk aw qosi ta)
If the continuous process is sampled at intervals ^x,
then the spectrum of the resulting discrete data will be :
| od
"iE chi
Pal«w) Ryj(k) © (4)
Ka SO
The inverse Fourier transform will give :
RU
\
Ra (&) = de Fa(w) cos Wk dw (5)
- 307 =