When w(r) can be expressed this way, the window is called a parameter window. 
3. Equivalent degree of freedom y, as the y° distribution that was v = 2 for the 
estimator Py(q@), changes into 
2N 2N 
0) a (2.98) 
loo) 
> wr(r) M {| Pwdu 
—-o 
4. The spectrum window blurs the real spectrum, Fig. 2.16. Its extent depends on 
the window bandwidth (BW) which is defined in several different ways, although every 
definition tries to show the equivalent frequency range over which the smoothing is per- 
formed. For a single-peaked spectrum, this blur depends on the relative bandwidths of 
the spectrum and the spectrum window. Spectrum bandwidth is defined as the frequency 
range in which the spectrum shows half of the peak of the spectrum. 
fee ae /\ 
1 7a Ne 4 
w, Ws > © te —| —~ w 
2n/M 
27/M 
Fig. 2.16. Blur of the spectrum by Daniel’s window, where the 
bandwidth is 27/M. 
(_ 
S 
Ze 
The resolution will be lower as the window bandwidth becomes narrower. 
2.5.5 Expression of Confidence Interval and Precision of §(w) 
1. By x2 Approximation 
36 
