Page 517 - Fundamentals of Radar Signal Processing

estimates in a manner similar to Eq. (6.129)

(6.152)

If an interfering target is present in one of the two windows it will raise the
interference power estimate in that window. Thus, the lesser of the two
estimates is more likely to be representative of the true interference level and
should be used to set the threshold.

Because the interference power is estimated from N/2 cells instead of N
cells, the threshold multiplier α required for a given design value of will be
increased. It is tempting to conclude that the threshold multiplier α for SOCA
SO
CFAR could be calculated using Eq. (6.135) with N replaced by N/2. A more
careful analysis shows that the required multiplier is the solution of the equation
(Weiss, 1982)

(6.153)

This equation must be solved iteratively. As an example, for and N =

20, α = 11.276. In contrast, the CA CFAR multiplier is α = 8.25 for the same
SO
conditions.
Figure 6.27a compares the behavior of conventional CA CFAR and SOCA
CFAR on simulated data containing two closely spaced targets of mean SNR 15
and 20 dB, a 10-dB clutter edge, and a third 15-dB target near the clutter edge.
As before, the lead and lag windows are both 10 samples (thus N = 20) and
there are three guard cells to each side of the test cell. The ideal threshold

shown is based on . The threshold multipliers are α = 8.25 for CA
CFAR and α = 11.276 for SOCA CFAR as discussed previously.
SO