Page 509 - Fundamentals of Radar Signal Processing

6.5.4 CA CFAR Limitations
The cell-averaging CFAR concept relies on two major assumptions:

1. Targets are isolated; specifically, targets are separated by at least the
reference window size, so that no two are ever in the reference window

at the same time.
2. All of the reference window interference samples are independent and
identically distributed, and that distribution is the same as that of the
interference component in the cell containing the target; in other words,
the interference is homogeneous.

While useful in many situations, either or both of these conditions are frequently
violated in real-world scenarios. The second assumption is particularly likely
to be untrue when the dominant interference is clutter, i.e., echo from terrain,
rather than thermal noise. In this section the effect on cell-averaging CFAR of
violating these assumptions is discussed, and then some modifications that
combat these effects are described.

Target masking occurs when two or more targets are present such that,
when one target is in the test cell, one or more targets are located among the
reference cells. Assuming that the power of the target in the reference cell
exceeds that of the surrounding interference, its presence will raise the estimate
of the interference power and thus of the CFAR threshold. The target(s) in the
reference window can “mask” the target in the test cell because the increased
threshold will cause a reduction in the probability of detection, i.e., the

detection is more likely to be missed. Equivalently, a higher SNR will be
required to achieve a specified .
Figure 6.22 is an example of target masking. As before, the interference
level is 20 dB, the target in range bin 50 has an SNR of 15 dB, and the threshold

–3
is computed using 20 reference cells and a desired of 10 . However, a
second target with an SNR of 20 dB in range bin 58 elevates the estimated
interference power when the first target is in the test cell. This increase in
threshold is sufficient to prevent detection of the first target in this case. On the
other hand, the 15 dB target does not affect the threshold enough to prevent
detection of the second, stronger target.