Window Functions and Spectral Leakage - Signal Processing of Random Physiological Signals - page 178

Weighting functions are applied three ways:

1. as a rectangular function, which does not modify the input data,

2. by having all the weighting function coefficients stored in memory, and

3. by computing each coefficient when it is needed.

16.2 PERFORMANCE MEASURES
The choice of a weighting function depends on which of the features of the narrowband
DFT filters ( resolution ) are most important to the application. Those features are perfor-
mance measures of the narrowband filters to analytically compare weighting functions.
All these measures, except frequency straddle loss, refer to individual filters. Frequency
straddle loss is associated with how filters work together [3].

16.2.1 Highest Side-Lobe Level
Highest side-lobe level (in decibels) is an indication on how large the effect of the side
lobes is. Side lobes are a way of describing how a filter responds to signals at frequencies
that are not in its main lobe, commonly called its passband. Each FFT filter has several
side lobes. With rare exception, the highest side lobe is closest in frequency to the main
lobe and is the one that is most likely to cause the passband filter to respond when it
should not. For a signal with the maximum point of the main lobe at 0 dB, rectangular
window the first side lobe is about

13 dB, relative amplitude. A good leakage suppressor
should have its first side lobe much lower than

−

13 dB. Since side lobes are the result
of leakage from the main lobe, then the more side lobes on either side of the main lobe
will result in a reduced or lower main lobe magnitude.

−

16.2.2 Sidelobe Fall-Off Ratio
The peak (amplitude) of side lobes decreases or remains level as they get further away in
frequency from the main lobe passband. Side lobe fall-off characteristic is very important
since it indicates the rate at which the side lobes are decreasing in magnitude. The