Digital Signal Processing Reference
In-Depth Information
Fig. 6.8
STOP curves for
MRE and HYCA approaches.
Note that for a practical range
of operating SNR values, the
HYCA based approach
consistently suggests a higher
P
FA
. Note that the STOP
curves suggest that the
desired false alarm
probability of the detector
should be readjusted
according to SNR variations.
This is different from the
conventional approach in
which the desired false alarm
probabilities are fixed
Each of these optimum operating line equations can be combined with the ROC
curve relation given in (
6.41
) to obtain a relation between
P
FA
and
ζ
as
P
1
/(
1
+
ζ)
FA
+
9
.
523
P
FA
−
1
.
002
=
0
for MRE,
(6.44)
P
1
/(
1
+
ζ)
FA
+
5
.
943
P
FA
−
0
.
979
=
0
for HYCA.
(6.45)
The numerical solution of these equations constitutes the
static threshold optimiza-
tion
(STOP) curves given in Fig.
6.8
. The resulting overall system target tracking
performance corresponding to the two static methods is also of interest and is in-
vestigated in detail in subsequent experimental sections. At this point, however, it
can be observed that for a practical range of operating SNR values, HYCA-based
optimization consistently suggests a higher
P
FA
, and therefore results in more false
detections for the radar processor to handle.
15
6.4.2 Experiment 1: Comparison with Heuristic Approaches
In this experiment, we compare the tracking systems given in Table
6.1
, each con-
sisting of a PDA tracking filter and a Neyman-Pearson (NP) front-end detector.
In the first three systems, the detectors use conventional (i.e., heuristically se-
lected) constant desired false alarm probabilities of
P
FA
=
10
−
8
,
P
FA
=
10
−
6
, and
10
−
4
, which are the typical values used in practice [
57
]. On the other hand,
the other systems utilize tracker-aware detectors for which the desired false alarm
P
FA
=
15
This is an important practical problem in the radar. Under excessive number of false detections,
the radar may initiate lots of false tracks. This causes the radar to allocate its resources, e.g., dwell
time, transmission power, unnecessarily and inefficiently.
Search WWH ::
Custom Search