Page 193 - Radar Technology Encyclopedia

183 filter, acousto-optical filter, autoregressive

the bounds of the aperture, the filter realizes a rectangular fre-
s(n) ^ y(n)
quency characteristic. 1
An acousto-optical filter can delay a signal, perform
-
band filtering, calculate the convolution of a signal with some S
reference, perform matched filtering, and so forth. e (n)
y(n) +
The basic filter characteristics range from 10 MHz to 2 to
3 GHz with a band up to 30%; number of resolved frequency
2
points depends on the time aperture of the acousto-optical
3
modulator, and is on the order of 10 ; dynamic range exceeds
Figure F22 Adaptive filter (after Cowan, 1985, Fig. 1.1, p. 11).
50 dB.
Acousto-optical filters are more difficult to produce than
other analog filters (CCD filters, SAW filters), so their use
Adaptive filters require a minimum volume of initial
becomes advisable for parallel processing of large masses of
information concerning the arriving signal. Gradient methods
data in real time. IAM
(see ALGORITHM, Widrow, and CANCELER, How-
Ref.: Kulikov (1989), p. 97; Zmuda (1994), p. 409.
ells-Applebaum) usually are used to control filter weight
An active filter is one that contains active elements (usually parameters. IIR and FIR filters with realization in both the
an amplifier) with R-C networks in its input and feedback time and the frequency domains are used as adaptive filters.
paths. An advantage of the active filter is that use of inductors Time to establish the adaptation mode, stability, and
may generally be avoided. The active elements are bipolar residual error are important adaptive filter characteristics.
and field-effect transistors, PIN diodes linked with the dielec- Adaptive filters are used widely in radar mainly as adaptive
tric cavity of a stripline, in the centimeter band or with a filter filter-compensators and filter-extrapolators. IAM
based on waveguide-slot lines in the millimeter band. Active Ref.: Cowan (1985); Widrow (1985), p. 18.
filers are used as bandpass or bandstop filters. The use of
An analog filter is one using analog circuit technology.
wideband active elements (field-effect transistors based on
Depending on the required filter response and its operating
gallium arsenide) in a filter with varactor frequency tuning
mode, an analog filter may be the simplest electronic circuit
makes it possible to obtain vanishingly small losses, along
or a complex device. Analog filters are divided into the fol-
with high tuning speed (less than 100 ns). DKB, IAM
lowing types: filters using oscillating or aperiodic circuits,
Ref.: Fink (1982), p. 12.34; Jordan (1985), Ch. 10.
using phase and dispersive circuits; and filters using delay
An adaptive filter is one that changes its parameters on the lines. Each type usually is used for filtering of a specific class
basis of current information to obtain a performance optimum of signals in matched processing and to obtain a different type
in a certain sense under initial uncertainty and changing oper- of frequency responses (see Table F2).The types of analog fil-
ating conditions. Input signal S(n) is filtered or weighted in an ters used in modern signal processing are acousto-optical,
adaptive filter (1) to obtain output signal y(n), which then is CCD, and SAW’ filters. IAM
compared with useful, standard, or training signal y n ()
to
antialiasing filter (see ALIASING).
find error signal e(n) (see Fig. F22). The error signal is used
with adaptation algorithm (2) for adjustment of the weighted An autoregressive filter is an all-pole filter in which parame-
factors of the filter parameters (usually using an integrative ters of an observed N-point data sequence are modeled as
method) for gradual minimization of the error (or of another those of white noise passed through a filter with N - 1 poles.
cost function). Ref.: Nitzberg (1992), p. 290.

Table F2
Analog Filter Types
Time-bandwidth
Filter type Type of filter implementation Area of use
product
Based on oscillating and LC, RC filters, quartz, piezocer- Filtering of coherent and B » 1
aperiodic circuits amic, electromechanical filters noncoherent narrowband
signals
Based on phase and dis- Cascaded S-filters, electro- Nonlinear filtering B » 10
persive circuits acoustic dispersive circuits, con-
nected in series or parallel
Based on delay lines Tapped delay lines Signal filtering with large B B » 100