Page 114 - Radar Technology Encyclopedia

104 CROWBAR data differentiation

Table C9
Crowbar Devices
Plasma
EG&G Energy
Hydrogen Mercury Multigap triggered
triggered gas Systems ball
thyratron pool ignitron crowbar vacuum
gap gap
gap
Voltage rating (kV, max) 40–125 50 70 No limit No limit 75
Current and joule rating 5–10 kA 50 kA 100 kA, 4 kJ No limit No limit 70 kA
Firing range 10:1 50:1 3:1 “infinite”* “infinite” “infinite”
Self-firing No No No No Yes No
Triggered end Negative Negative Either Either Either Negative
Trigger V as a fraction of voltage 1/10 1/50 1/3 1/2 1/2 5 kV (all
rating sizes)
Size Small Small Small Large Large Small
Cost Low Low Medium Medium High Medium
*With some inductance in series with load, which limits effectiveness.
(from Skolnik, 1990, Table 4.4, p. 4.40, reprinted by permission of McGraw-Hill).

D

DATA, radar. The term radar data generally refers to the bors) association. In the first case, the total summed distance
information gathered about the target by a radar or a set of from observations (reports) to the assigned tracks is mini-
radars. Typically, this information makes possible the deter- mized. The state estimate (position and velocity in each coor-
mination of radar target coordinates (range, azimuth, eleva- dinate) for each track is projected forward to the next scan,
tion, radial velocity) and reflective characteristics of the target and when the signal processor provides the next set of reports
(e.g., radar cross section). The radar data are collected, pro- the algorithm assigns to each track the plot that is closest to
cessed, stored, and displayed in the radar channel. Various its prediction. In the all-neighbors approach the association
information measures can be used for theoretical description problem is considered a matrix, each row representing an
of the quantity and quality of information circulating in the existing track and each column a number of reports for the
radar channel. SAL current scan. The ijth element of the matrix indicates the like-
Ref.: Skolnik (1962), p. 453; Tuchkov (1985), p. 11. lihood (or probability) that the jth report is associated with the
ith track. The optimum solution for the report-to-track assign-
Data association is the procedure of assigning a set of esti-
ment requires selecting the set of report-track pairs that maxi-
mate of the dynamic states of the targets (e.g. tracks) to a set
mizes the sum of the matrix entries. The nearest neighbor
of measurements generated by a radar or a set of radars. This
approach is much simpler, but target-tracking performance
is sometimes referred to as the correlation problem, which is
can be severely degraded by misassociation. The all-neigh-
a serious problem in data fusion and multiple-target tracking
bors approach provides considerable improvement in associa-
tasks. In this case, a decision must be made about the corre-
tion performance but at the expense of complexity, often
spondence of a given set of measurements with a given set of
requiring considerable computation resources. SAL
tracks and estimates made of correctness of the decision. Typ-
ically, these estimates are the probability of correct associa- Ref.: Hovanessian (1988), p. 253; Bar-Shalom (1992), v. II, pp. 183–227.
tion and the probability of misassociation. These probabilities Data differentiation is part of the process of radar measure-
depend primarily on the ratio between radar resolution (accu- ment. Differentiating is used primarily to determine the target
racy) and the density of the target within radar coverage. The velocity using derivatives obtained from radar position data
simple analytic models for predicting association perfor- output. Velocity in three coordinates can be obtained from a
mance can be derived only for some specific cases. In the single radar only by differentiation. The process is performed
general case, a computer simulation using the Monte Carlo by differentiating circuits or algorithms, typically results in an
approach is the most appropriate. increase in noise, and must always be accompanied by
The two basic algorithms used to implement data associ- smoothing. SAL
ation are nearest neighbor and matrix assignment (all-neigh- Ref.: Barton (1964), pp. 422–428.