Page 535 - Fundamentals of Radar Signal Processing
P. 535
(6.183)
Equation (6.183) illustrates another limitation of this technique: small values of
require large values of N. More realistically, practical limitations on the
reference window size N limit this method to relatively high values of . For a
given value of N, the design value of the probability of false alarm, say ,
must be chosen to be greater than or equal to . Assuming this condition is
satisfied, the rank order to be used as a threshold is the one that produces a
value of as close to as possible without exceeding it. That rank is given
by
(6.184)
While the false alarm probability does not depend on the PDF of the
interference, the detection probability does. For exponentially distributed
interference and target (Swerling 1 target in complex WGN), the average
probability of detection is (Sarma and Tufts, 2001)
(6.185)
As with the other CFAR detectors, Eqs. (6.184) and (6.185) can be used to
determine the CFAR loss of the DF CFAR. The additional loss over a CA
CFAR for the Swerling 1 case is typically less than about 0.4 dB.
6.6 System-Level Control of False Alarms
It has been seen in this chapter that achieving good detection performance (high
, low ) requires a signal-to-interference ratio on the order of 15 dB or
better at the point of detection. For a given target RCS, the SIR is determined in
part by basic radar system design choices reflected in the radar range equation:
transmitter power, antenna gain, operating frequency, and noise figure.
Furthermore, the fundamental goal of many of the techniques of radar signal
processing discussed in other chapters of this text is to improve the SIR before
the point of detection. Examples include matched filtering, pulse compression,
MTI, pulse Doppler processing, and space-time adaptive processing. Once the
SIR has been maximized, the detector, whether fixed or adaptive threshold, sets
the actual threshold value and thus the false alarm probability. The SIR then
determines the detection probability.

