 |
Claims  |
|
|
I claim:
1. A method for converting a filtering a plurality of input signals in a
sensor system into one filtered output signal, said method comprising the
steps of:
obtaining a first main input signal, and a plurality of first auxiliary
input signals,
determining interior weights corresponding to said first main and auxiliary
input signals,
determining equivalent linear weights corresponding to said interior
weights,
obtaining a second main input signal and a plurality of second auxiliary
input signals, said second main and auxiliary signals being obtained in
substantially the same noise environment as said first main and auxiliary
signals, and
adding said second main signal to the product of said equivalent linear
weights and said second auxiliary input signals to thereby produce one
filtered output signal corresponding to said second main and auxiliary
signals, and
wherein said first main and auxiliary input signals are a subset of said
second main and auxiliary input signals, respectively.
2. The method of claim 1, wherein said first main and auxiliary input
signals are equal to said second main and auxiliary input signals,
respectively.
3. An apparatus for converting and filtering a plurality of input signals
in a sensor system into one filtered output signal, said apparatus
comprises:
a first main input channel for receiving a first main input signal from a
main sensor,
a plurality of first auxiliary input channels for receiving first
auxilliary input signals from a plurality of auxiliary sensors,
Gram-Schmidt processor means, connected to said first main and auxiliary
input channels, for determining interior weights corresponding to said
first main and auxiliary input signals,
calculator means connected to said Gram-Schmidt processor means for
receiving said interior weights from said Gram-Schmidt processor means and
for calculating the equivalent linear weights corresponding to said
interior weights,
a second main input channel for receiving a second main input signal from
said main sensor,
a plurality of second auxiliary input channels for receiving second
auxiliary input signals from said auxiliary sensors,
canceller means, connected to said calculator means to receive said
equivalent linear weights and connected to said second main and auxilliary
input channels to receive said second main and auxiliary input signals,
for adding said second main signal to the product of said equivalent
linear weights and said second auxiliary signals to produce one filtered
output signal corresponding to said second main and auxiliary input
signals, and
wherein said first main and auxiliary input signals are a subset of said
second main and auxiliary input signals, respectively.
4. The apparatus of claim 3, wherein said sensor system comprises a moving
target indicator system and said main radar sensor and said plurality of
auxiliary radar sensors comprise radar antennas for receiving radar echo
signals.
5. The apparatus of claim 3, wherein said first main and auxiliary input
signals are equal to said second main and auxiliary input signals,
respectively.
6. A method for converting and filtering a plurality of input signals in a
sensor system into one filtered output signal, said method comprising the
steps of:
obtaining a first main input signal, and a plurality of first auxiliary
input signals,
determining interior weights corresponding to said first main and auxiliary
input signals, wherein said interior weights satisfy the equation:
w.sub.n.sup.(m) (j)=x.sub.N-m.sup.(m)t (j) x.sub.n.sup.(m)
(j)/x.sub.N-m.sup.(m)t (j) x.sub.N-m.sup.(m) (j),
where
n=0,1, . . . , N-m-1
x.sub.n.sup.(m+1) (j)=x.sub.n.sup.(m) (j)-w.sub.n.sup.(m) (j)
x.sub.N-m.sup.(m) (j),
m=1,2, . . . , N-1
`t` denotes the conjugate vector transpose operation, x.sub.0.sup.(1) (j)
denotes said main input signal, x.sub.1.sup.(1) (j), x.sub.2.sup.(1) (j),
. . . , x.sub.N-1.sup.(1) (j), denote said auxilliary input signals, "j"
is a time index and N is an integer greater than 1,
determining equivalent linear weights corresponding to said interior
weights, wherein said equivalent linear weights satisfy the equations:
W.sub.1 =w.sub.0.sup.(N-1)
W.sub.2 =w.sub.0.sup.(N-2) -w.sub.1.sup.(N-2) W.sub.1
W.sub.3 =w.sub.0.sup.(N-3) -w.sub.1.sup.(N-3) W.sub.1 +w.sub.2.sup.(N-3)
W.sub.2
W.sub.N-1 =w.sub.0.sup.(1) -w.sub.1.sup.(1) W.sub.1 +w.sub.2.sup.(1)
W.sub.2 - . . . (-1).sup.N-2 w.sub.N-2.sup.(1) W.sub.N-2.
obtaining a second main input signal and a plurality of second auxiliary
input signals wherein said second main and auxiliary signals are obtained
in substantially the same noise environment as said first main and
auxiliary signals,
producing one filtered output signal z, wherein z satisfies the equation:
z=y.sub.0 -W.sub.1 y.sub.1 -W.sub.2 y.sub.2 . . . W.sub.N-1 y.sub.N-1
and y.sub.0 denotes said second main input signal and y.sub.1, y.sub.2, . .
. , y.sub.N-1 denote said second auxiliary input signals.
7. The method of claim 6, further comprising storing said equivalent linear
weights in a memory.
8. The method of claim 6, wherein said first main and auxiliary input
signals are equal to said second main and auxiliary input signals,
respectively.
9. The method of claim 6, wherein said first main and auxiliary input
signals are a subset of said second main and auxiliary input signals,
respectively.
10. An apparatus for converting and filtering a plurality of input signals
in a sensor system into one filtered output signal, said apparatus
comprises:
a first main input channel for receiving a first main input signal from a
main sensor,
a plurality of first auxiliary input channels for receiving first auxiliary
input signals from a plurality of auxiliary sensors,
Gram-Schmidt processor means, connected to said main and auxiliary input
channels, for determining interior weights corresponding to said main and
auxiliary input signals, wherein said interior weights satisfy the
equation:
w.sub.n.sup.(m) (j)=x.sub.N-m.sup.(m)t (j) x.sub.n.sup.(m)
(j)/x.sub.N-m.sup.(m)t (j) x.sub.N-m.sup.(m) (j),
where
n=0,1, . . . , N-m-1
x.sub.n.sup.(m+1) (j)=x.sub.n.sup.(m) (j)-w.sub.n.sup.(m) (j)
x.sub.N-m.sup.(m) (j),
m=1,2, . . . , N-1
`t` denotes the conjugate vector transpose operation, x.sub.0.sup.(1) (j)
denotes said main input signal, x.sub.1.sup.(1) (j), x.sub.2.sup.(1) (j),
. . . , X.sub.N-1.sup.(1) (j), denote said auxiliary input signals, "j" is
a time index and N is an integer greater than 1,
calculator means connected to said Gram-Schmidt processor means for
receiving said interior weights from said Gram-Schmidt processor means and
for calculating the equivalent linear weights corresponding to said
interior weights, wherein said equivalent linear weights satisfy the
equation:
##EQU4##
a second main input channel for receiving a second main input signal from
said main sensor,
a plurality of second auxiliary input channels for receiving second
auxiliary input signals from said auxiliary sensors,
canceller means, connected to said calculator means to receive said
equivalent linear weights, and connected to said second main and auxiliary
input channels to receive said second main and auxiliary input signals,
for producing one filtered output signal z, wherein z satisfies the
equation:
z=y.sub.0 -W.sub.1 y.sub.1 -W.sub.2 y.sub.2 . . . W.sub.N-1 y.sub.N-1
and y.sub.0 denotes said second main input signal and y.sub.1, y.sub.2, . .
. , y.sub.N-1 denote said second auxiliary input signals.
11. The apparatus of claim 10, wherein said sensor system comprises a
moving target indicator system and said main radar sensor and said
plurality of auxiliary radar sensors comprise radar antennas for receiving
radar echo signals.
12. The apparatus of claim 10, wherein said first main and auxiliary input
signals are equal to said second main and auxiliary input signals,
respectively.
13. The apparatus of claim 10, wherein said first main and auxiliary input
signals are a subset of said second main and auxiliary input signals,
respectively. |
|
 |
|
 |
Claims |
|
|
|
Description |
|
|
FIELD OF THE INVENTION
The invention is directed to signal processing systems and, in particular,
to signal processing systems employing an adaptive noise canceller with a
slaved Gram Schmidt processor.
BACKGROUND OF THE INVENTION
Signal information received by a radar system frequently includes unwanted
echoes reflected from stationary or slowly moving reflectors such as the
ground or sea, or from wind driven rain or chaff. The unwanted echoes
obscure desired signals--such as those reflected from a moving target. The
desired signals corresponding to a moving target usually vary quickly with
time. The unwanted signals corresponding to sea clutter and wind driven
rain or chaff vary slowly with time. This difference can be exploited to
eliminate the unwanted signals since data signals corresponding to quickly
varying signals are uncorrelated as a function of time whereas slowly
varying signals are correlated with time. In other words, stationary
reflectors yield return echoes that include frequency components which
vary more slowly than the frequency components of the desired moving
target. Therefore, unwanted signals are reduced or eliminated by
separating the received signals into correlated and uncorrelated
components and then by filtering out the correlated components.
To this end, various systems and techniques have been devised which filter
unwanted signals by isolating the correlated components of a received
signal and then canceling the correlated components from the received
signal. One such system is a Gram-Schmidt canceller which receives a main
signal and a plurality of auxiliary signals all corresponding to the same
target object. The Gram-Schmidt canceller correlates each auxiliary signal
with the main signal, generates weighting factors commensurate with the
degree of correlation of the signals, and uses the weighting factors to
eliminate the correlated components from the main signal to thereby yield
one filtered output signal. Each additional data set is processed by the
Gram-Schmidt canceller and new weighting factors are generated appropriate
to the new data set. Such conventional systems are described in more
detail below when comparing these systems with the present invention.
Systems using a Gram-Schmidt canceller or a similar technique also include
those disclosed in the following references: U.S. Statutory Invention H92
(Kretschmer et al) which discloses a moving target indicator system
incorporating a Gram Schmidt processor for reducing clutter in the signal;
U.S. Statutory Invention H14 (Lewis) which discloses a moving target
indicator system including an adaptive noise canceller using
sliding-window derived weights; U.S. Pat. No. 4,688,187 (McWhirter) which
discloses a Gram-Schmidt algorithm for applying linear constraints
(weighting factors) to data inputs; U.S. Pat. No. 4,652,881 which
discloses a moving target indicator system using a Gram-Schmidt adaptive
noise canceller to decorrelate auxiliary signals from a main signal; and
U.S. Pat. No. 4,398,197 (Dillard) which discloses a digital sidelobe
canceller using weighted coefficients in the noise reduction process.
SUMMARY OF THE INVENTION
In accordance with one aspect of the invention a method is provided for
converting and filtering a plurality of input signals in a sensor system
into one filtered output signal, the method comprising the steps of
obtaining a first main input signal and a plurality of first auxiliary
input signals, determining interior weights corresponding the first main
and auxiliary signals, and determining equivalent linear weights
corresponding to the interior weights. A second main input signal and a
plurality of second auxiliary input signals are also obtained with the
second main and auxiliary input signals being obtained in substantially
the same noise environment as the first main and auxiliary input signals.
The second main input signal is added to the product of the equivalent
linear weights and the second auxiliary input signals to produce one
filtered output signal corresponding to the second main and auxiliary
signals.
In accordance with a further aspect of the invention, an apparatus is
provided for filtering signals in a sensor system, wherein the apparatus
includes: a main input channel for receiving first main input signals from
a main sensor, a plurality of auxiliary input channels for receiving first
auxiliary input signals from a plurality of auxiliary sensors, a
Gram-Schmidt processor, connected to the first main and auxiliary input
channels, for determining interior weights corresponding to the first main
and auxiliary input signals, calculator means for calculating equivalent
linear weights corresponding to the interior weights, and canceller means
connected to the calculator means for receiving the equivalent linear
weights, and connected to second main and auxiliary input channels for
receiving second main and auxiliary input signals, wherein the second main
and auxiliary input signals are obtained in substantially the same noise
environment as the first main and auxiliary input signals, and wherein the
canceller means adds the second main signal to the product of the
equivalent linear weights and the second auxiliary signals, to produce a
single filtered output signal corresponding to the second main and
auxiliary signals.
As will be evident from the foregoing and will be discussed in more detail
below, the invention is a modified Gram-Schmidt processor for use with a
signal processing system having a plurality of sensors each receiving
signal input corresponding to the same target or source. The signal
processing system can comprise a conventional system such as a radar
moving target indicator system, a sonar system, or a communication system.
However, the invention can be advantageously applied to any signal
receiving system which includes a plurality of signal receiving means.
To better understand how a conventional Gram-Schmidt processor is modified
in accordance with the invention, it is helpful to consider the operation
of such a conventional processor. As noted above, a conventional
Gram-Schmidt processor receives input data signals along one main channel
and a plurality of auxiliary channels. The data signals along each of the
channels include desired signals generated within the same noise
environment. The desired signals are assumed to be uncorrelated between
data channels whereas at least a portion of the noise is correlated
between data channels. A conventional Gram-Schmidt processor correlates
the auxiliary data channels with the main channels and generates interior
weights corresponding to the correlated components. The processor then
applies the interior weights to the signals and eliminates the correlated
components from the main signal to thereby eliminate correlated noise
components.
Considering an exemplary application, in a typical moving target indication
system, a plurality of radar antennas are employed to track one target and
the data signals received along the various data channels all include
dynamic signals corresponding to the same target. The signals also include
relatively stationary noise components corresponding to sea clutter and
wind driven rain or chaff. The dynamic target signals are typically
uncorrelated between channels because the targets are quickly moving,
whereas the slowly varying signals are relatively uncorrelated between
signals. Hence, a conventional Gram-Schmidt processor can be
advantageously used to eliminate the clutter from the moving target
indication system signals. However, because of a relatively long
processing time, the conventional Gram-Schmidt processor is ill-suited to
large data sets. Moreover, the interior weights, once calculated, are
discarded because the interior weights generated by the Gram-Schmidt
processor can not be applied to any other data signal.
The present invention involves a modification of the conventional
Gram-Schmidt processor wherein reusable weighting factors are generated.
The reusable weighting factors, referred to as "equivalent linear
weights," are generated from an initial set of signals, and thereafter are
used to filter noise from other signals having similar stationary noise
components. Furthermore, for processing large data sets, equivalent linear
weights can be calculated from a relatively small subset of the data set,
then applied to the entire data set to thereby quickly filter the entire
data set. Thus, a system constructed in accordance with the invention is
capable of quickly filtering data sets which are too large to be
efficiently processed by a conventional Gram-Schmidt processor. In
general, the equivalent linear weights can be applied to any new signals
generated in the same noise environment.
Returning to the example of a moving target indication system, similar
noise components appear in all signals received from the system. Since the
noise environment is relatively stationary, equivalent linear weights,
once generated, can be re-used to quickly filter the ground and sea
clutter from new signals. It will be appreciated from the foregoing that a
key advantage of the invention is in the generation of equivalent linear
weights which can be applied to additional data sets. Once an appropriate
set of equivalent linear weights is generated for a particular noise
environment, the weights can be reused for other data sets which include
the same or similar noise components.
Another advantage of the invention is in the numerically efficient
implementation of the calculator means which uses systolic processing to
generate equivalent linear weights. Similar hardware is used to implement
both the Gram Schmidt processor means and the calculator means and
therefore the invention is also hardware efficient.
Other features and advantages will be set forth in, or be apparent from,
the detailed description of the preferred embodiments which follows.
BRIEF DESCRIPTION OF THE DRAWINGS
FIG. 1 is a block diagram of an adaptive processor construction in
accordance with the preferred embodiment of the invention.
FIG. 2(a) is a block diagram of the Gram-Schmidt processor of the
invention.
FIG. 2(b) is a block diagram of the two input Gram-Schmidt canceller of the
invention.
FIG. 3 is a block diagram of an exemplary embodiment of the Gram-Schmidt
processor having three auxiliary input lines.
FIG. 4a is a block diagram of an exemplary embodiment of the equivalent
linear weight calculator of the invention.
FIG. 4b is a block diagram of the weighted subtractor of the invention.
FIG. 5 is a block diagram of an alternative embodiment of the adaptive
processor of FIG. 1.
FIG. 6 is a schematic representation of target overflying terrain including
water and land mass and illustrates an application of the invention as
part of a multiple-radar moving target indicator system.
DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS
Referring to FIGS. 1-4, a preferred embodiment of the invention will now be
described. Referring first to FIG. 1, input to the adaptive processor is
received along main input data line 10 and along a plurality of auxiliary
input data lines 12. The signals received along main input line 10 and
auxiliary input lines 12 include quickly varying target signals having
slowly varying or stationary noise components. Main input line 10 includes
preprocessor 14 for processing the main input signal to produce two
separate signals which appear on two datalines 10a and 10b. In particular,
10a carries only a subset of the data provided along data line 10.
Likewise auxiliary input data lines are processed via preprocessor 16 into
separate data lines 12a and 12b with 12a carrying a subset of the data
provided along dataline 12. Preprocessors 12 and 14 are of conventional
construction.
Main input line 10a and auxiliary input lines 12a are connected to a
systolic Gram Schmidt processor 18 which generates interior weights
corresponding to correlated stationary noise components of the main and
auxiliary signals. By calculating the interior weights, systolic
Gram-Schmidt processor 18 also generates a filtered output signal based on
the data received along datalines 10a and 12a. This filtered output signal
can optionally be output along a filtered output dataline (not shown).
In the preferred embodiment of FIG. 1, only the interior weights are output
from the systolic Gram-Schmidt processor 18. The form of the interior
weights and the details of systolic Gram Schmidt processor 18 are
described below in connection with the description of FIGS. 2a and 2b. The
interior weights are output from systolic Gram-Schmidt processor 18
through data line 20 to an equivalent linear weight calculator 22 which
produces re-usable equivalent linear weights corresponding to the interior
weights provided by Gram Schmidt processor 18.
The equivalent linear weights can subsequently be applied to any data set
having the same noise components as the data provided along datalines 10a
and 10b and therefore can be applied to the entire input data signals
provided on datalines 10b and 12b. The form of the equivalent weights and
the details of equivalent linear weight calculator 22 are described below
in connection with the description of FIG. 4. Together systolic
Gram-Schmidt processor 18 and equivalent linear weight calculator 22
comprise what is referred to as a slaved Gram-Schmidt processor and is
denoted 24 in FIG. 1.
The equivalent weights produced by equivalent linear weight calculator 22
are output along data lines 26 to a canceller 28 which also receives the
entire main and auxiliary input signals along datalines 10b and 12b.
Canceller 28 weights the auxiliary input signals from dataline 12b by the
equivalent linear weights and combines the weighted auxiliary input
signals with the main input signal from dataline 10b to produce a filtered
output signal which is output along output data line 30. The function of
canceller 28 is detailed below.
Thus, the entire dataset received along input datalines 10 and 12 is
filtered using equivalent linear weights calculated from only a subset of
the original dataset. Therefore, in use, if an input dataset is too large
to be efficiently filtered by systolic Gram-Schmidt processor 18,
splitters 14 and 16 are set to allow, for example, a subset of one out of
every five data points to be passed into datalines 10a and 10b, such that
equivalent linear weights are calculated from one fifth of the total
dataset, then applied to filter the entire dataset.
Referring to FIGS. 2(a) and 2(b), a detailed description of the systolic
Gram-Schmidt processor 18 will now be provided. A block diagram of
Gram-Schmidt processor 10 is provided in FIG. 2(a) wherein x.sub.0,
x.sub.1, . . . , x.sub.N-1 represent the complex data in the 0th, 1st, . .
. , N-1th channels, respectively. The main signal channel input of
dataline 10a is represented by x.sub.0 and the remaining N-1 inputs are
the auxiliary channels input on dataline 12a.
The main channel signal consists of a desired signal along with stationary
noise signals. Cancellation of the stationary noise signals is achieved by
correlating the simultaneously received signals in the main and auxiliary
channels. Noise which is uncorrelated between channels is not filtered by
this system.
Gram Schmidt processor 18 operates to individually decorrelate each
auxiliary input from all other inputs by using a plurality of two-input
Gram-Schmidt cancellers 32 arranged in a hierarchy of levels. One such
two-input Gram-Schmidt canceller 32 is shown schematically in FIG. 2(b),
and will described in more detail below. As shown in FIG. 2(a), x.sub.N-1
is decorrelated with x.sub.0, x.sub.1, . . . x.sub.N-2 in the first level
of decomposition yielding x.sub.0.sup.(2), x.sub.1.sup.(2), . . .
x.sub.N-2.sup.(2). Next, x.sub.N-2.sup.(2) is decorrelated with
x.sub.0.sup.(2), . . . x.sub.1.sup.(2), . . . x.sub.n-3.sup.(2). This
decorrelation process continues until one output channel remains
(x.sub.0.sup.(N)). However, the output channel is not totally decorrelated
with the input because the interior weights in each of the two-input
Gram-Schmidt cancellers 32 are computed from a finite number of input
samples rather than an infinite number. Thus, the interior weights are
generated by this technique are estimates of optimal interior weights.
The systolic Gram-Schmidt processor 18 processes data on a point by point
basis, with interior weights estimated at each time step. The interior
weights are calculated according to a fixed number of previous samples at
any given point in time. For each time step the latest sample is included
and the oldest sample is discarded.
The operation of systolic Gram-Schmidt processor 18 is described
mathematically as follows. For an indexed time instant j,
x.sub.n (j)=[x.sub.n (j),x.sub.n (j-1), . . . x.sub.n (j-K+1)].sup.T,n=0,
1, 2, . . . N-1 (1)
where K is the number of samples used to calculate each of the interior
weights and T denotes the vector transpose operation. Gram Schmidt
cancellers 32 are arranged in N-1 levels, each receiving input from the
previous level as shown in FIG. 2(a). The input vectors of an mth level of
Gram-Schmidt canceller 32 are x.sub.n.sup.(m) (j) and x.sub.n+1.sup.(m)
(j), where
x.sub.n.sup.(m) (j)=[x.sub.n.sup.(m) (j), x.sub.n.sup.(m) (j-1), . . .
x.sub.n.sup.(m) (j-k+1)].sup.T, (2)
where x.sub.n.sup.(1) (j)=x.sub.n (j) and x.sub.n.sup.(1) (j)=x.sub.n (j).
The output of each two-input Gram-Schmidt canceller 32 at the (m+1) level
is:
##EQU1##
`t` denotes the conjugate vector transpose operation, and w.sub.n.sup.(m)
(j) is the interior weight.
The interior weights are not time aligned estimates because of the effect
of systolic processing, rather the interior weights at any level of
systolic Gram-Schmidt processor 18 are one time step ahead of the interior
weights at a successive level.
FIG. 3 provides a schematic diagram of an exemplary embodiment of systolic
Gram-Schmidt processor 18 with N=4 wherein for clarity only the internal
weights are shown. Time delay, T, is the systolic step time The term
w.sub.n.sup.(m) represents the steady state value of w.sub.n.sup.(m) (j)
corresponding to an infinite number of samples per channel, and hence
w.sub.n.sup.(m) represents the optimal weight. In prior art Gram-Schmidt
processor these interior weights are not saved, rather the interior
weights are merely used as intermediate values for generating a filtered
output signal. Hence, in prior art systems, the interior weights are
discarded after each level of the Gram-Schmidt processor. Here, however,
the interior weights are not discarded but are transferred along dataline
26 to equivalent linear weight calculator 22 for conversion into
equivalent linear weights.
Referring to FIG. 4, the equivalent linear weight calculator will now be
described. For each set of interior weights, equivalent linear weights,
W.sub.1, W.sub.2, . . . , W.sub.N-1, corresponding to the interior weights
are generated as follows:
##EQU2##
Because the interior weights calculated by systolic Gram-Schmidt processor
18 are not time-aligned, the equivalent linear weights are computed by
using time delayed estimates of the interior weights, or
##EQU3##
FIG. 4a shows a functional block diagram of an exemplary embodiment of
equivalent linear weight calculator 22 with N=4. Similarly to systolic
Gram-Schmidt processor 18, the equivalent linear weight calculator 22
works on a clocked basis where data is calculated and transferred at a
clock rate denoted by T. The equivalent linear weight calculator 22
implements the equations for the equivalent linear weights given in
Equation (5). Gram-Schmidt interior weights at time instant j,
w.sub.n.sup.(m) (j), are input into the equivalent linear weight
calculator 22. These inputs are appropriately time delayed by
.tau..sub.n.sup.(m), where:
.tau..sub.n.sup.(m) =(N-1-m+n)T (7)
Equivalent linear weight calculator 22 includes a plurality of weighted
subtractors 50. As shown in FIG. 4b with exemplary input "a" "b" and "w"
input "b" to each weighted subtractor is weighted by w and subtracted from
input "a".
In an alternative embodiment (not shown) of the equivalent linear weight
calculator 22, the first column of weighted subtractors 50 shown in FIG.
4a are not used, rather the output of the .tau..sub.n.sup.(m) delay is
directly connected to the T delay. This achieves the same result as the
embodiment of FIG. 4a since the input to each first column weighted
subtractor 50 is 0 and -1, and therefore the output of each first column
weighted subtractor is identical to the input weight.
Returning to the preferred embodiment of FIG. 1, the equivalent linear
weights W.sub.1, W.sub.2, . . . , W.sub.N-1 are transferred along dataline
26 to canceller 28 and applied to the auxiliary signals. The entire main
and auxiliary input datasets are also received by canceller 28 along
datalines 10b and 12b respectively. Canceller 28 generates a filtered
output signal, z, by numerically solving the equation:
z=x.sub.0 -W.sub.1 x.sub.1 -W.sub.2 x.sub.2 . . . W.sub.N-1 x.sub.N-1(8)
To solve equation (8), canceller 28 multiplies each auxiliary data point by
the corresponding equivalent linear weight then subtracts the net result
from the main dataline points, yielding the filtered signal output along
dataline 30.
Thus, the preferred embodiment of FIGS. 1-4 generates one filtered output
signal from a plurality of input signals by first calculating
noise-filtering weighting factors from a subset of the input signals, then
applying the weighting factors to the entire input signals.
A block diagram of an alternative embodiment of the invention is provided
in FIG. 5. The embodiment of FIG. 5 is similar to the embodiment of FIG.
1, with the addition of a memory 32 connected to canceller 28 and the
provision for separate input lines 10a, 10b, 12a and 12b without
preprocessors 14 and 16. As described above in the summary of the
invention, equivalent linear weights from an initial dataset, once
calculated, can be applied to any new datasets having the same noise
components as the initial dataset. The embodiment of FIG. 6 exploits this
capability.
The embodiment of FIG. 5 uses memory 32, which are of conventional design,
in storing the equivalent linear weights received along dataline 26. In
use, an initial dataset is input along datalines 10a and 10b and
equivalent linear weights are generated corresponding thereto. New
datasets, having the same or similar noise components as the initial set,
are received along datalines 10b and 12b. The stored equivalent weights
are then used to filter the new datasets. Cancellation is accomplished in
the same manner as described above for FIG. 1.
Optionally, memory 32 can be used to store a plurality of equivalent linear
weight sets corresponding to a plurality of noise environments. A selector
circuit, not shown, can be provided to choose a particular set of
equivalent linear weights to filter signals received in the corresponding
noise environment.
FIG. 6 shows an exemplary system in which the invention can be
advantageously employed. A multiple-radar moving target indicator system
50 includes an adaptive noise filter 60 constructed in accordance with
either the preferred or alternative embodiments as described above. A main
input channel 56 and a plurality of auxiliary input channels 58 carry
signals received from a main radar antenna 52 and a plurality of auxiliary
antennas 54, respectively, to adaptive noise filter 60. The moving target
indicator system 50 further includes a signal processing and display unit
66 which is of conventional design and will not be described further.
Radar returns from target 62 are partially obscured by radar echoes
received from sea surface 64. Adaptive noise filter 60 operates, as
described above, to eliminate the unwanted sea clutter radar echoes to
thereby yield a noise-free signal corresponding to target 62.
Although the invention has been described with respect to exemplary
embodiments thereof, it will be understood by those skilled in the art
that variations and modifications can be effected in these exemplary
embodiments without departing from the scope and spirit of the invention.
* * * * *
|
|
|
|
|
|
|
Description |
|
