Page 373 - Fundamentals of Radar Signal Processing

(5.78)

To determine the window’s effect on the noise power, suppose y[m] is a zero
2
mean stationary white noise with variance σ . Then the windowed noise power
is

(5.79)

The unwindowed noise power N can be obtained from Eq. (5.79) by setting
2
w[m] = 1 for all m, giving N = Mσ . Using Eq. (5.77) and these values for N and
N in Eq. (5.79) gives the processing loss
w

(5.80)

Like the loss in peak gain, the processing loss is a weak function of M that

is higher for small M but quickly approaches an asymptotic value. As an
example, for the Hamming window the loss in SNR is 1.75 dB for a very short
(M = 8) window, decreasing asymptotically to about 1.35 dB for long windows.
Table 5.2 summarizes the four key properties of 3-dB resolution, peak sidelobe
level, loss in processing gain, and processing loss for several common
windows. Also shown is worst-case straddle loss, to be discussed next. All of
these characteristics of windows are the same as discussed in Chap. 4 in the
context of range side-lobe control. A much more extensive table, including both

more metrics and many more types of windows, is given in Harris (1978). 8