Page 102 - Radar Technology Encyclopedia

92 CFAR, cell-averaging (CACFAR) CFAR, hierarchal

quency to the target filter. In pulsed doppler radar, range- clutter visibility inherent in the use of a high-resolution
gated filters surrounding the target cell in both range and dop- search radar in non-Rayleigh clutter. Output of the clutter
pler may be used (Fig. C36). map may also be used to control the input dynamic range of
the receiver, to select processing paths appropriate to each
Reference cells
type of interference, and to blank cells containing interference
From
receiver so strong that no other processing can prevent false alarms.
Integrator
Clutter-map CFAR is often called temporal [time-averaging]
CFAR or simply a clutter map. DKB
Detections
C Ref.: Skolnik (1990); Nitzberg (1992), pp. 233–236.
S Dicke-fix CFAR is a cell-averaging CFAR technique in the
X frequency domain, in which a broadband IF amplifier is fol-
lowed by a limiter and a narrowband amplifier (Fig. C37).
K
Figure C35 Cell-averaging CFAR (after Skolnik, 1990,
Fig. 8.12, p. 8.13). From Wideband IF Narrowband To
IF limiter IF
Doppler mixer amplifier amplifier detector
frequency
Bandwidth, B w Bandwidth, B n
Figure C37 Dicke-fix CFAR.
Target detection
cell
The effective number of reference cells is equal to the band-
Range
w
n
delay width ratio, M = B /B . When used with pulse compression,
the number of cells becomes BtM, making CFAR possible
Reference with M = 1 (see phase-discrimination CFAR). DKB
cells
Ref.: Skolnik (1980), p. 394.
Figure C36 Use of reference cells from combined range
distribution-free CFAR (see nonparametric CFAR).
and doppler regions surrounding the target cell.
Greater-of cell-averaging CFAR is a technique in which the
The CFAR constant K is set to produce the desired false-
reference cells are divided in leading and lagging cells, with
alarm probability in the output of the threshold detector C.
separate averages being taken (Fig. C38). The greater of the
The performance of cell-averaging CFAR depends on the
number of reference cells and their extent in the selected
Reference cells
radar coordinate. A CFAR loss is introduced by the fact that
the estimate of interference level in the reference cells is sub- From
receiver Integrator
ject to random error, which varies inversely with the number
of cells, forcing the user to use K values higher than would be
Detections
used if the estimate were exact. On the other hand, if the num- C
ber of cells is increased to reduce this loss, the circuit fails to
S C S
respond to rapid changes in interference level, permitting a X
burst of false alarms to occur when this level increases rap-
K
idly. Also, the greater the number of reference cells the
Figure C38 Greater-of cell-averaging CFAR.
greater the probability that an adjacent target will fall into the
reference cells, and distort the threshold setting (see multi-
two averages is used to control the threshold, permitting more
ple-target CFAR). Use of combined range and doppler
rapid adaptivity to changes in the interference environment at
regions around the target cell is one means of avoiding exces-
some expense in CFAR loss. The increase in CFAR loss for
sive delay in CFAR response in either coordinate. DKB
this configuration, with respect to conventional cell-averag-
Ref.: Skolnik (1990); Nitzberg (1992), Ch. 9.
ing CFAR, is typically about 0.2 dB. DKB
A clutter-map CFAR sets the detection threshold in a given
Ref.: Nitzberg (1992), pp. 226–229.
test cell based on the average interference observed in that
Hierarchal CFAR techniques perform separate calculations
cell (and often several adjacent cells) over several scans of
of the threshold using two methods (e.g., CACFAR with
the antenna. The time average may be determined recursively,
parameters chosen for an environment of noise, and an adap-
using an a-b filter. For fluctuating interference, the time aver-
tive alternative suitable for clutter or jamming). The normal
aging increases the precision of the interference estimate
CACFAR threshold T is used, with minimum CFAR loss,
without extending the spatial extent of the reference region.
unless the second method yields a threshold, T > aT, where
By reducing the size of the reference region (in the limit to a
a > 1. DKB
the test cell itself), the clutter-map CFAR preserves the inter-
Ref.: Nitzberg (1992), pp. 245–249.