Page 124 - Radar Technology Encyclopedia

114 detection, accumulator detection, binary

Accumulator detection uses an accumulator (an adder with a In automatic detection the decision about target presence is
memory) containing a delay element and amplification logic done by electronic circuitry without the involvement of a
(Fig. D6). The amplification logic assigns a weight a to each radar operator. The basic aspects of automatic detection
signal crossing the first threshold T , and a weight b to each include:
1
failed detection. To make a decision about the presence of the (1) Integration of received pulses when more than one is
target, the instantaneous sum at the accumulator output is expected to be received.
compared with a second threshold T . SAL (2) Decision-making as to target presence.
2
Ref.: Neri (1991), p. 119. Frequently the means of maintaining a constant false-alarm
rate (CFAR) are also included in automatic detection cir-
X + a
cuitry, in which case the detector is called a CFAR detector.
Accumulator
Sampler + U The functions of target detection are often combined with
- + + +
-
measurement of target location in range and angular coordi-
Th 1
X - b Delay nates (e.g., azimuth). In this case the automatic detector is
First threshold (1 PRI)
called a plot extractor or data extractor. SAL, AIL
Th 2
U Second threshold Ref.: Skolnik (1970), Ch. 15, (1980), p. 388; Kuz’min (1974), pp. 5–10;
N a Schleher (1980).
High SNR
PRI Automatic detection and tracking is the automatization of
the target detection and tracking process when the functions
Th 2
Low SNR of target detection, track initiation, track association, track
t
update, track smoothing, and track termination are performed
Figure D6 Accumulator detection (after Neri, 1991, Fig. 2.55, in an automatic mode. SAL
p. 120). Ref.: Skolnik (1980), p.183.
In adaptive detection the parameters and even the structure The Bayes criterion of detection is the optimum detection
of a detector can change depending on the signal and interfer- criterion producing the minimum average risk when the deci-
ence levels, to maintain some performance characteristics at a sion as to target detection is made under the conditions of
specified level or to respond in a specified way to variation in known a priori probability of a signal presence.
input data. An example of an adaptive detector is the CFAR To apply the criterion, one has to know a priori probabil-
receiver, when such a parameter as the probability of false ity P of the fact that a signal u(t) is present in the received
a
alarm is maintained at the constant level depending on inter- realization y(t), and the average cost for risk r has to be
ference environment characteristics. The block diagram of the assigned. If the cost of a false alarm is r , and the cost of
fa
simplest adaptive detector is cited at Fig. D7. This detector missing the target is r , then in general, when many detec-
tm
tions are performed during radar operation, the average cost
IF input Video Threshold
IF amp
fa a a
detector unit of false alarms is r q P , and the average cost of missing tar-
Output decision
a
tm a tm
fa
tm
a
on presence or gets is r P P , where, q = 1 - P , and P , P are the
Gain-control absence of signal probabilities of false alarm and missing the target, respec-
voltage
Differential Low-pass tively. The best detector under this criterion will be that which
IF amp filter
requires minimum average risk for both erroneous decisions:
r = r q P + r P P = min.
Reference voltage fa a fa tm a tm
The intelligent choice of the coefficients r and r tm is not
fa
Figure D7 Simplified adaptive detection. a trivial task and is based on the assumption of how danger-
can adapt to slow fluctuations of noise level by means of an ous and undesirable each kind of error is. The Bayes criterion
IF amplifier with an automatic gain control circuit. The is generic and many other detection criteria are special cases
abrupt increase in level because of the signal presence, makes of it. Another term for this criterion is the minimum average
the threshold unit work before the AGC circuit will make up risk criterion. AIL
for the level change. More complicated self-adjustable adap- Ref.: Barkat (1991), p. 118; Kazarinov (1990), p. 24
tive detectors combined with the adaptation of surveillance
In binary detection a decision is made between one of two
period and waveform structures are typically used in phased
possible outcomes: noise alone or signal plus noise. The deci-
array radars. AIL
sion is made only about the existence of a target, and target
Ref.: Skolnik (1970), pp. 15–29; Shirman (1981), pp. 313–316; Sosulin parameters such as range, velocity, and others are not
(1978), p. 292-306.
obtained without further processing. In a binary detection
In a posteriori detection a decision is made on the basis that problem, it is typically considered that the transmitter may
the estimated amplitude will be larger than some minimum send a deterministic signal s (t) under the null hypothesis H ,
0
0
value chosen for reliable detection. This approach is closely or deterministic signal s (t) under the alternative hypothesis
1
related to maximum likelihood detection. SAL H . When the signal is received, it is corrupted with an addi-
1
Ref.: DiFranco (1980), pp. 229–233. tive noise process (white or colored), and the goal is to decide