Methods and systems for locating targets using non linear radar with a matched filter which uses exponential value of the transmit signal

0Associated
Cases 
0Associated
Defendants 
0Accused
Products 
0Forward
Citations 
0
Petitions 
5
Assignments
First Claim
1. A method of nonlinear radar target location comprising:
 transmitting, with an transmitting antenna, a signal of a transmit waveform towards a target;
receiving, with a receiving antenna, a signal from the target; and
by at least one processor;
creating a matched filter by generating an exponential function of the transmit waveform corresponding to a particular harmonic of the interest; and
applying the matched filter to the received signal to generate and output a signature waveform for the target of the particular harmonic of interest.
5 Assignments
0 Petitions
Accused Products
Abstract
Embodiments of the present invention concern locating targets using nonlinear radar with a matched filter which uses exponential value of the transmit signal. According to embodiments, a method of nonlinear radar target location includes: transmitting a signal of a transmit waveform towards a target; receiving a signal from the target; creating a matched filter by generating an exponential function of the transmit waveform corresponding to a particular harmonic of the interest; and applying the matched filter to the received signal to generate and output a signature waveform for the target of the particular harmonic of interest. In other embodiments, the matched filtering may be combined with sidelobe reduction.
25 Citations
No References
RF receiver sensing by harmonic generation  
Patent #
US 7,864,107 B1
Filed 06/12/2009

Current Assignee
Rockwell Collins Inc.

Sponsoring Entity
Rockwell Collins Inc.

Explosive device countermeasures  
Patent #
US 7,987,068 B2
Filed 03/25/2009

Current Assignee
Curators of the University of Missouri

Sponsoring Entity
Curators of the University of Missouri

PHASE MATCHING BANDPASS FILTER USING EXPONENTIAL FUNCTION APPROXIMATION  
Patent #
US 20110291751A1
Filed 01/22/2010

Current Assignee
Samsung Electronics Co. Ltd.

Sponsoring Entity
Samsung Electronics Co. Ltd.

DISMOUNT HARMONIC ACCELERATION MATCHED FILTERING FOR ENHANCED DETECTION AND DISCRIMINATION  
Patent #
US 20100245152A1
Filed 03/31/2009

Current Assignee
Raytheon Company

Sponsoring Entity
Raytheon Company

Radar system and method  
Patent #
US 7,777,671 B2
Filed 07/12/2005

Current Assignee
RAFAELARMAMENT DEVELOPMENT AUTHORTIY LTD.

Sponsoring Entity
RAFAELARMAMENT DEVELOPMENT AUTHORTIY LTD.

Method and system for forming an image with enhanced contrast and/or reduced noise  
Patent #
US 7,796,829 B2
Filed 12/10/2008

Current Assignee
The United States of America As Represented By The Secretary of Agriculture

Sponsoring Entity
United States Of America As Represented By The Secretary Of The Army

Semiconductor article harmonic identification  
Patent #
US 6,856,275 B1
Filed 12/26/2001

Current Assignee
Raytheon Company

Sponsoring Entity
Raytheon Company

Nonlinear junction detector  
Patent #
US 6,897,777 B2
Filed 02/11/2002

Current Assignee
AUDIOTEL INTERNATIONAL LIMITED

Sponsoring Entity
AUDIOTEL INTERNATIONAL LIMITED

Radio frequency detection and identification system  
Patent #
US 6,894,614 B2
Filed 05/04/2001

Current Assignee
CHECKPOINT SYSTEMS INC.

Sponsoring Entity
CHECKPOINT SYSTEMS INC.

System and method of radar detection of nonlinear interfaces  
Patent #
US 6,765,527 B2
Filed 01/31/2002

Current Assignee
Johns Hopkins University

Sponsoring Entity
Johns Hopkins University

Surveillance apparatus and method for the detection of radio receivers  
Patent #
US 6,049,301 A
Filed 09/22/1976

Current Assignee
The Boeing Co.

Sponsoring Entity
The Boeing Co.

Frequency mixing passive transponder  
Patent #
US 6,060,815 A
Filed 08/18/1997

Current Assignee
XCyte Inc.

Sponsoring Entity
XCyte Inc.

Pulse transmitting nonlinear junction detector  
Patent #
US 6,163,259 A
Filed 06/04/1999

Current Assignee
RESEARCH ELECTRONICS INTERNATIONAL

Sponsoring Entity
RESEARCH ELECTRONICS INTERNATIONAL

Apparatus and method for pulse compression and pulse generation  
Patent #
US 5,557,560 A
Filed 05/24/1994

Current Assignee
SECRETARY OF STATE FOR DEFENCE IN HER BRITANNIC MAJESTYS

Sponsoring Entity
SECRETARY OF STATE FOR DEFENCE IN HER BRITANNIC MAJESTYS

Radar target signature detector  
Patent #
US 5,191,343 A
Filed 02/10/1992

Current Assignee
WESTINGHOUSE NORDEN SYSTEMS INCORPORATED

Sponsoring Entity
United Technologies Corporation

Radar object detector using nonlinearities  
Patent #
US 4,053,891 A
Filed 05/24/1967

Current Assignee
Lockheed Electronics Company

Sponsoring Entity
Lockheed Electronics Company

Metal target detection system  
Patent #
US 3,972,042 A
Filed 12/02/1974

Current Assignee
Motorola Inc.

Sponsoring Entity
Motorola Inc.

Method and apparatus for remote detection of radiofrequency devices  
Patent #
US 8,131,239 B1
Filed 08/21/2007

Current Assignee
Vadum Inc.

Sponsoring Entity
Vadum Inc.

System and method for iterative fourier side lobe reduction  
Patent #
US 8,665,132 B2
Filed 03/11/2011

Current Assignee
United States Of America As Represented By The Secretary Of The Army

Sponsoring Entity
United States Of America As Represented By The Secretary Of The Army

JUNCTION RANGE FINDER  
Patent #
US 3,732,567 A
Filed 09/21/1970

Current Assignee
Na, Low George M Deputy Administrator of The National Aeronautics and Space, Sarto Morissette, Ronald G. Sea, Marvin J. Frazier

Sponsoring Entity
Na, Low George M Deputy Administrator of The National Aeronautics and Space, Sarto Morissette, Ronald G. Sea, Marvin J. Frazier

HARMONIC RADAR DETECTING AND RANGING SYSTEM FOR AUTOMOTIVE VEHICLES  
Patent #
US 3,781,879 A
Filed 06/30/1972

Current Assignee
RCA Corporation

Sponsoring Entity
RCA Corporation

COMBINED RADAR ASSEMBLY WITH LINEAR AND NONLINEAR RADAR  
Patent #
US 20150084811A1
Filed 09/20/2013

Current Assignee
United States Of America As Represented By The Secretary Of The Army

Sponsoring Entity
United States Of America As Represented By The Secretary Of The Army

Multitone Harmonic Radar and Method of Use  
Patent #
US 20150253415A1
Filed 04/25/2013

Current Assignee
United States Of America As Represented By The Secretary Of The Army

Sponsoring Entity
United States Of America As Represented By The Secretary Of The Army

Target detection utilizing image array comparison  
Patent #
US 9,250,323 B2
Filed 05/24/2012

Current Assignee
ARMY UNITED STATES GOVERNMENT AS REPRESENTED BY THE SECRETARY OF THE

Sponsoring Entity
ARMY UNITED STATES GOVERNMENT AS REPRESENTED BY THE SECRETARY OF THE

Clutter suppression in ultrasonic imaging systems  
Patent #
US 9,451,932 B2
Filed 02/29/2012

Current Assignee
Crystalview Medical Imaging Limited

Sponsoring Entity
Crystalview Medical Imaging Limited

20 Claims
 1. A method of nonlinear radar target location comprising:
transmitting, with an transmitting antenna, a signal of a transmit waveform towards a target; receiving, with a receiving antenna, a signal from the target; and by at least one processor; creating a matched filter by generating an exponential function of the transmit waveform corresponding to a particular harmonic of the interest; and applying the matched filter to the received signal to generate and output a signature waveform for the target of the particular harmonic of interest.  View Dependent Claims (2, 3, 4, 5, 6, 7, 8, 9, 10, 11, 12, 13, 14, 15, 16, 17)
 18. A system for nonlinear radar target location comprising:
a transmitter having an antenna which is configured to transmit a signal of a transmit waveform towards a target; a receiver having an antenna which is configured to receive a signal from the target; and at least one processor which is configured to; create a matched filter by generating an exponential function of the transmit waveform corresponding to a particular harmonic of the interest; and apply the matched filter to the received signal to generate and output a signature waveform for the target of the particular harmonic of interest.  View Dependent Claims (19)
 20. A method of nonlinear radar target location comprising:
transmitting, with an transmitting antenna, a signal of a transmit waveform towards a target; receiving, with a receiving antenna, a signal from the target; and by at least one processor; analyzing the transform waveform to determine if it is real or complex; and calculating an analytic signal version of the transmit waveform for subsequent use as the transmit waveform, if the transmit waveform is determined to be real; creating a matched filter by generating an exponential function of the transmit waveform corresponding to a particular harmonic of the interest; filtering the received signal with the matched filter signal; performing a recursive sidelobe minimization process (RSM) in a plurality of iterations, wherein each iteration of the RSM process comprises; randomly excising a subset of frequency samples from within a frequency band of the matched filtered received signal; and reducing sidelobe levels of the matched filtered received signal by a windowing operation.
1 Specification
Governmental Interest—The invention described herein may be manufactured, used and licensed by or for the U.S. Government.
A computer program listing appendix has been submitted via EFSWeb labeled as “codeappendix.” The material contained in the appendix is incorporated by reference herein as though rewritten fully herein.
i) Field of Invention
This application generally relates to radar, and more particularly, to nonlinear radar and processing steps, which exploit nonlinear target responses.
ii) Description of Related Art
Historically, radar systems have operated on signals that are assumed to be echoes produced by linear systems. In linear radar systems, a pulse transmitted at one frequency will reflect off an object and produce a receive signal at the same frequency. This paradigm generally functions well, enabling radar systems to detect targets provided that their signatures are sufficiently different from those of natural clutter objects. Unfortunately, it has been demonstrated that certain clutter objects produce large linear responses to radar probe signals. For example, trees and rocks are capable reflecting relatively large amounts of energy, making it difficult for radar operators to distinguish them from targets of interest.
Advances in radar technology have made the problem somewhat more tractable. Modern synthetic aperture radar (SAR) systems are capable of achieving high resolution in both downrange and crossrange. While these systems produce detailed imagery of a scene, the detection of smaller targets can still be problematic. For example, targets that are only a few inches in diameter may produce a radar signature that is indistinguishable from the clutter background. Modern systems that exploit Doppler phenomena face a similar problem. In this case, large returns from stationary clutter objects can obscure slowmoving targets and, especially, if the targets reflect only a small amount of energy relative to the nearby clutter. It is the width of the main clutter lobe that helps fix the minimum detectable velocity achieved by a moving target indication (MTI) system.
Improvements in nonlinear radar would be useful.
Embodiments of the present invention concern methods and systems for locating targets using nonlinear radar with a matched filter which uses exponential value of the transmit signal.
According to embodiments, a method of nonlinear radar target location comprises: transmitting, with an transmitting antenna, a signal of a transmit waveform towards a target; receiving, with a receiving antenna, a signal from the target; and by way of at least one processor: creating a matched filter by generating an exponential function of the transmit waveform corresponding to a particular harmonic of the interest; and applying the matched filter to the received signal to generate and output a signature waveform for the target of the particular harmonic of interest.
Before the step of creating the matched filter, the method may further comprise analyzing the transform waveform to determine if it is real or complex; and calculating an analytic signal version of the transmit waveform for subsequent use as the transmit waveform, if the transmit waveform is determined to be real. Calculating the analytic signal version of the transmit waveform may be achieved by applying a Hilbert transform to the transmit waveform, for instance.
The particular harmonic of the interest can be characterized as n*f_{1}, where n≥2 and f_{1 }is the fundamental frequency of the transmit waveform and the exponential function may be defined as ƒ^{n }(n). Thus, in some instances, the harmonic of interest is a second harmonic and the exponential function is the square of the transmit waveform. Or the harmonic of interest is a third harmonic and the exponential function is the cube of the transmit waveform. Higher harmonics are also envisioned.
In general, the exponential function is assumed to be an expected target response signature such that the output of the matched filter is a correlation between the received signal and the expected target response signature. The transmit waveform may be any waveform suitable for radar applications, such as, a chirp waveform, pseudorandom noise waveform, polyphase code waveform, Frank code waveform, Barker code waveform, or pseudorandom code waveform.
To apply the matched filter to the received signal, the methodology can include: sampling the received signal over a period of time; timereversing the exponential function generally corresponding to the same period of time; and comparing the sampled received signal and timereversed the exponential function via a convolution process to determine the time delay or phase shift between the two signals.
The convolution process may comprise populating a received signal vector with sampled signal data of the received signal, and a target impulse response vector with signal data corresponding to the exponential function. If there are any unpopulated data fields in the received signal or the target impulse response vectors, zero values may be added in their place. The convolution process may be thought of as an iterative process, in which each iteration involves: multiplying corresponding vector index values of the received signal and the target impulse response vectors together; summing the multiplied values; shifting the target impulse vector values one index position to the right; and wrapping rightmost shiftedposition value to the leftmostposition in the vector. After completing all iterations, the maximum summed value for all the iterations is identified. A new received vector is generated for a new sampling of the received signal, and the target impulse response vector values stay unchanged so long as the same transmit waveform is used. A complex conjugate function may be used to execute these steps, for example.
In further processing, in some embodiments, sidelobe reduction is applied in combination with the matched filtering. The combined sidelobe reduction and matched filtering can include: filtering the received signal with the matched filter; and performing a recursive sidelobe minimization process (RSM) in a plurality of iterations, wherein each iteration of the RSM process comprises: randomly excising a subset of frequency samples from within a frequency band of the matched filtered received signal; and reducing sidelobe levels of said signal by a windowing operation. The received signal and the matched filter could be both transformed into the frequency domain prior to the filtering operation for simplified processing compared to in the time domain. The windowing operation might include a Hanning window, a Taylor window, or a rectangular window, for instance.
According other embodiments, a system for nonlinear radar target location includes: a transmitter having an antenna which is configured to transmit a signal of a transmit waveform towards a target; a receiver having an antenna which is configured to receive a signal from the target; and at least one processor which is configured to: create a matched filter by generating an exponential function of the transmit waveform corresponding to a particular harmonic of the interest; and apply the matched filter to the received signal to generate and output a signature waveform for the target of the particular harmonic of interest. The transmitter includes a wave generator for generating to transmit waveform.
According to further embodiments, a method of nonlinear radar target location includes: transmitting, with an transmitting antenna, a signal of a transmit waveform towards a target; receiving, with a receiving antenna, a signal from the target; and by way of at least one processor: analyzing the transform waveform to determine if it is real or complex; and calculating an analytic signal version of the transmit waveform for subsequent use as the transmit waveform, if the transmit waveform is determined to be real; creating a matched filter by generating an exponential function of the transmit waveform corresponding to a particular harmonic of the interest; filtering the received signal with the matched filter signal; performing a recursive sidelobe minimization process (RSM) in a plurality of iterations, wherein each iteration of the RSM process comprises: randomly excising a subset of frequency samples from within a frequency band of the matched filtered received signal; and reducing sidelobe levels of the matched filtered received signal by a windowing operation.
These and other embodiments are further discussed below.
So that the manner in which the above recited features of the present invention can be understood in detail, a more particular description of the invention, briefly summarized above, may be had by reference to embodiments, some of which are illustrated in the appended drawings. It is to be noted, however, that the appended drawings illustrate only a few embodiments of this invention and are therefore not to be considered limiting of its scope, for the invention may admit to other equally effective embodiments. These embodiments are intended to be included within the following description and protected by the accompanying claims.
Embodiments of the present invention provide improved nonlinear detection and ranging capabilities. It has been demonstrated that certain objects produce nonlinear responses to radar probe signals. Nonlinear radar produces frequencies in a nonlinear target (e.g., electronics or metal object) response that are different from those transmitted by the radar, thereby separating natural clutter from the nonlinear target response. In particular, echoes can be observed at one or more harmonics of the transmitted frequencies. Thus, the processing described herein does not focus on the entire transmit waveform, which contains multiple discrete signal tones, but instead it focuses on its harmonic(s)ofinterest.
The novel methodology detects and processes harmonic target responses to a transmitted radar waveform, and in this way, it is able to localize the target in range.
The transmit waveform or waveforms should be wellunderstood. Typical, such signals are periodic function. That is: If the frequencydomain representation of the transmitted signal during a time interval t_{i }is: then the target response (i.e. the time series) will be: Σ_{n }a_{n,i }cos(ωt +ϕ_{i})^{n}, where i denotes time intervals (which are disjoint), n is the power series index, and a_{n,i }is the power series coefficient. Preferably, a 2^{nd }harmonic target response is used as that signal is typically the strongest harmonic. Of course, higher harmonics (e.g, 3^{rd}, 4^{th}, 5^{th }. . . n^{th}) may also be considered. This follows from the fact that the nonlinear response can be represented as a power series, and different terms in the series produce different harmonics.
To isolate and detect a particular nonlinear harmonic of interest, a matched filter is utilized which uses an exponential function of the transmit waveform signal. The exponential is whole number greater than or equal to 2. This corresponds to the square, cube, quadric, etc. The matched filter correlates a parameter of a known signal with an unknown signal to detect the presence of the parameter in the unknown signal. In this case, the parameter of interest is frequency or period associated with a particular harmonic. It turns out the overall frequency of the exponential functions are the same as their corresponding harmonics. This matched filter is then used to enable and/or improve detection of targets which are expected to generate a response at this specific harmonic. The matched filter attempts to match the received signal to an expected version of the target response for a particular harmonic, based on the corresponding exponential of the transmit signal. Advantageously, by using this methodology, the fundamental frequency of the transmit signal, which the harmonic frequencies are multiples of, does not need to be known or calculated.
For the power series product term squared (n=2), cos^{2 }(θ_{1}) is equal to ½+½ cos(2θ_{i}), where θ_{i}=ω_{i}t+ϕ_{i}. This can be mathematical shown by way of a trigonometric identity. The first component here is a DC component, whereas the second component is a cosine function of double the fundamental frequency. When summed this second component dictates the frequency of the squared function.
Similarly, for the power series product term cubed (n=3), cos^{3}(θ_{i}) is equal to ¾ cos(θ_{i})+¼ cos(3θ_{i}). The first component here corresponds to the fundamental frequency, whereas the second component is a cosine function of triple the fundamental frequency. When summed this second component dictates the overall frequency of the cubic function.
And, for the power series product term to the fourth power (n=4), cos^{4}(θ_{i}) is equal to ⅜+½ cos(2θ_{i})+⅛ cos (4θ_{i}). The first component here is a DC component, the second term is a cosine function of double the fundamental frequency and the third component is a cosine component of quadruple frequency. When summed, this third component dictates the overall frequency of the fourth power function.
From the general pattern apparent here from the second to fourth harmonics, it should be becoming clear that there is harmonic component corresponding to the power of the function which dictates the overall frequency of the exponential function. Higher power of cosines can be deduced using De Moivre'"'"'s formula, Euler'"'"'s formula, binomial theorem, and/or other trigonometric identities. While the exponential may, in theory, be any whole number greater than or equal to 2, in actuality, the exponential of preferred embodiments is 2 or 3 and maybe 4. That is because beyond about the third or fourth power, the signal to noise ratio (SNR) may become too small to isolate the corresponding harmonic signal which becomes exponential smaller for each increasing power (½ times for n=2, ¼ times for n=3 and ⅛ times for n=4, etc.). Although a cosine function was described for explanation purposes above, it is believed that this phenomenon occurs with any periodic or repeating function, which may be characterized as combinations of sine and/or cosine functions, as in a Fourier series.
The matched filter exploits this mathematical phenomenon, by restricting attention of the received signal to the frequency of the exponential power of the transmitted signal. As shown above, this corresponds to a particular nonlinear harmonic frequency. So if we are interested in examining the second harmonic, we would use the frequency of the square of the transmitted signal for locating a second harmonic signal in the received signal data. And if we are interested in examining the third harmonic, we would use the frequency of the cube of the analytic form of the transmitted signal for locating a third harmonic signal in the received signal data, and so forth, using higher powered exponents for the higher corresponding harmonics.
By using the matched filter with an exponential function, we have more flexibility in the selection of our transmitted waveform. For instance, by using an analytic (i.e., a complex, Hilberttransformed) version of the signal, we eliminate the DC component in the squared version and effectively limit our consideration to the secondorder response. This is emphasized by the plot of
As explained herein, a novel matched filter 40 is connected to the transmitter 20 and the receiver 30. While shown adjacent to the transmitter 20 and receiver 30, it will be appreciated that the matched filter 40 could be located remotely from one or both of these elements. In other instance, it could be incorporated therein. The nonlinear radar system 10 can be mounted on an aircraft or a vehicle configured with transmitting and receiving antennas to transmit and measure the reflected radar signals from a target 50 of interest.
The radar system 10 may configured to transmit and receive in the 300 MHz to 3 GHz range, for instance. Since the target responses are assumed to be nonlinear, the methodology concentrates on the second and higher harmonics of the transmitted frequencies. The highest target returns are typically found at the second harmonics which are likely of primary interest. While one could use the third, fourth, or even higher harmonics, there may be lower efficiencies in terms of the received signal strength. It has been found that some targets (theoretically) generate only odd harmonics. This may be the result of metaltometal junctions, for example; so in these case, the odd harmonics would be of primary interest. Other physical phenomena and target interaction may produce other responses. The selected set of harmonics/intermodulation (intermod) products selected should be contiguous. Gaps in the spectrum of the received signal are to be avoided, since they introduce sidelobe artifacts.
In the transmitter 20, a transmit waveform signal is created by a waveform generator 21 which could be any conventional signal generator suitable for radar. The transmit waveform may be a periodic function of a single frequency tone f. Integer multiples of the original frequency (e.g. 2f, 3f 4f . . . ) are harmonics. The time domain representation of the transmitted waveform is denoted ƒ(t) in the continuous time domain or ƒ(n) in the discrete time domain. As most radar systems are digitally implemented, attention will primarily be paid to the discrete time domain. An amplifier 24 boots the transmit signal strength as would be suitable for the radar application. If desired, a filter, such as a lowpass filter 25 may be used to eliminate high frequencies and/or noise generated during amplification from the signal prior to transmission. The amplified signal is transmitted by the transmit antenna 27.
Radar systems process linear data produced by both the target 50 and natural clutter echoes. Certain targets of interest, however, produce nonlinear responses, while clutter objects do not. The target 50 may be an electronic device, such as a cellular phone as shown. Additional targets might include the metaltometal junctions on certain manmade objects, indicating the presence manmade devices; for example, metallic unexploded ordnance or tightlypacked, small metallic objects.
According to the novel nonlinear processing, by considering second or higher harmonic responses to our transmitted waveform, the processing is able to eliminate the often large responses due to natural clutter, such as trees and foliage. More particularly, by creating a matched filter 40 from exponentials of the transmitted signal, this nonlinear processing technique can adapt and extend the concepts of linear systems.
The backscattered radar signals from the target 50 or other imaging area along the radar path captured by the receiver 30. For instance, in the receiver 30, signals are received by a receiving antenna 32. The received signal is then passed through a lowpass filter 33 to eliminate higherfrequency data that is not of interest. It is then amplified and digitized, via amplifier 34 and digitizer 35, respectively, before being input to the detector 36, which performs any preliminary processing prior to the matched filter. Thus, a digitized, timedomain signal for the received signal may be input to the matched filter 40.
The matched filter 40 is configured to generate and output a signature waveform for the target signature of the desired harmonic. Ideally, a highresolution target signature. It may be implemented as machineexecutable or computerexecutableinstruction (e.g., software code) executed by a computer processing module, which is comprised of memory module(s) 42 and one or more processor(s) 44 (or microprocessors) as known in the art that are configurable to execute the novel processing methodology. Instructions 45, such as software code, firmware, or the like, may be stored on a computer or machinereadable storage media having computer or machineexecutable instructions executable by the processor(s) 44 configured them for executing of the processor(s). Processorexecutable instructions 45 can be stored in a nonvolatile memory device and executed by the processor(s) when needed. (
In some implementations, the processor(s) 44 may be a programmable processor, such as, for example, a fieldprogrammable gate array (FPGA) or an applicationspecific integrated circuit (ASIC) processor. The methodology disclosed herein may be implemented and executed by an application that may be created using any number of programming languages. An embodiment of invention has been verified using measured data and Mathworks® Matlab code. A copy of the Matlab code is incorporated by reference in the attached Appendix. Of course, any number of hardware implementations, programming languages, and operating platforms may be used without departing from the spirit or scope of the invention. As such, the description or recitation of any specific hardware implementation, programming language, and operating platform herein is exemplary only and should not be viewed as limiting. The methodology disclosed herein may be implemented and executed by an application may be created using any number of programming languages.
Alternatively, embodiments of the matched filtering could be implemented as hardware (e.g., electrical circuit) using delay lines, stored replicas of the waveform (e.g. an arbitrary waveform generator), etc. The key point of this disclosure, however, is the processing methodology. Of course, any number of hardware implementations, programming languages, and operating platforms may be used without departing from the spirit or scope of the invention. As such, the description or recitation of any specific hardware implementation, programming language, and operating platform herein is exemplary only and should not be viewed as limiting.
Immediately prior to or during transmission, a copy of the transmit waveform ƒ(n) is provided to the matched filter 40 from the antenna receiver 20; it may be generated by the waveform generator 21. The transmitted waveform ƒ(n) may be any waveform which can be used for radar communication. The n here in parenthesis represents the discrete time domain. Alternatively, the nomenclature t could be used to represent the continuous time domain and known in the art. In some implementations, the transmitted waveform may be characterized as a function, in the frequency domain, as F(n) where n here is a given frequency value or bin. (As in common parlance in the art, functions denoted in capital letters denote frequency domain versions of functions, i.e., Fourier transforms). Some common waveforms could include phasecoded waveforms, and steppedfrequency waveforms, for example.
In Step S10, a copy of the transmit waveform ƒ(n) is stored in memory. The copy may be stored in an electronic memory device which a processor accesses. Alternatively or additionally, various values of the transmit waveform ƒ(n) could be stored for values of n in a lookup table or database.
Next, in Step S20, the transmit waveform is further analyzed to see and if it should be converted to an analytic (complex) representation. This is to simplify processing as is generally appreciated in the art. For example, if the signal, ƒ(n) is real (meaning no complex data), then we calculate a corresponding analytic signal, ƒ_{A}(n). One way to do this is to use a Hilbert transform. Thus, we let ƒ_{A}(n)=H{ƒ(n)}, where H{ƒ(n)} is the Hilbert transform of ƒ(n). The Hilbert transform is important in signal processing, where it derives the analytic representation, in the complex plane (e.g., Inphase/Quadrature data is in polar form), of an input real signal. In Matlab, the function hilbert( ) returns a complex analytic signal, from a real data sequence. The analytic signal xr+i*xi has a real part, xr, which is the original data, and an imaginary part, xi, which contains the Hilbert transform. By performing a Hilbert transform, we can create the analytic (complex) representation of the received or transmitted signal: Σ_{n}a_{n,i}e^{−nj(ω}^{i}^{t}^{i}^{+ϕ}^{i}) where j=√{square root over (−1)}. This is a convenient simplification for the purposes of analysis and processing. If both inphase and quadrature measurements are available, though, then the Hilbert transform is not needed. We assume ƒ(n) is already complex here so no further mention of ƒ_{A}(n) will be made in the subsequent steps, but it will be appreciated the ƒ_{A}(n) would be used in lieu of ƒ(n).
In Step S30, a desired target harmonic of interest is determined. In some instances, the desired harmonic may be determined ahead of time, or perhaps by default. In other instances, a radar operator could input a target harmonic. The harmonic is a multiple of the fundamental frequency of the transmit waveform, n*f_{1}, where n≥2.
Next, in Step S40, an exponential function for the transmit waveform ƒ(n) is generated which corresponds to the desired target harmonic of interest in generated. The exponential function is generally defined as ƒ^{n }(n). Here, the exponent ^{n }will be the same as multiple n of the fundamental frequency of the transmit waveform from Step S30 corresponding to a desired target harmonic of interest. Any analytical function processor (•)^{n }can be used for generating the exponential function. Matlab provides such a capability. For a second harmonic, the exponential power is 2, and ƒ^{2}(n) is the squared transmitted function. The square of the transmit waveform, ƒ^{2}(n) is mathematically defined as ƒ(n)×ƒ(n). Thus, the matched filter in the case will be configured to look for the frequency of ƒ^{2}(n) which correspond to the second harmonic response in the received signal. Analogously, for a third harmonic, the exponential power is 3 and ƒ^{3}(n) is the cubed transmitted function. The cube of the transmit waveform, ƒ^{3}(n) is mathematically defined as ƒ(n)×ƒ(n)×ƒ(n). The matched filter here will be is configured to look for the frequency of ƒ^{3}(n) which correspond to the third harmonic response in the received signal.
In Step S50, the exponential function of the transmit waveform is applied to the received signal so as to identify and isolated the desired target harmonic. The output of the matched filter is ƒ_{m}(n) corresponds to a high resolution target signature of the desired harmonic. Using the exponential function at the expected analytic signal of the target signature, it is timereversed to obtain the matched filter. This timereversal is necessary so that the filter output becomes the correlation between the received signal and shifted versions of the expected target response. More particularly, the target impulse response of the matched filter has been configured to correlate with the expected target response present in the received signal r(n). The received signal r(n) may be received from the detector 36 of the receiver 30. The target impulse response may include the fundamental and various harmonics of the transmit waveform ƒ(n). But, for nonlinear radar application, we are usually interested in harmonics and more particular, in only one particular harmonic.
A convolution function can be used for correlation purposes in some instances. In particular, circular convolution is applicable since periodic functions are at play. For the convolution function, the target impulse response of the matched filter correlates with the expected target response. As discussed above, the exponential function of the transmit waveform ƒ(n) has the same frequency as the nth harmonic. Thus, if we focus on same frequency, we can limit ourselves to just that harmonic in the received signal r(n). In turns out this correlation can be explained, mathematical, with the matched filter being the timereversed conjugated version of the expected target response.
The following processing can make use of vector data structure. The vector structure may be a simple onedimensional array structure of the type used by various computer programming systems for storing data values in memory and processing. The length of the vectors are each N, which can be twice the length of the transmit waveform ƒ(n). Each data field or slot of the vectors is associated with an index location for ease of identifying and processing data in that particular field or slot. The fields or slots in the vectors correspond to small time steps (e.g., 1 ms of less). The indices of the vector are identified below for ease of explanation.
While the vector are illustrated as being horizontal, there is no limitation on their spatial arrangement; these data structures are merely illustrative of exemplary data processing. The vectors could be implemented as memory registers in some implementation. Two similar vectors are primarily used: (i) a received signal vector, (ii) a target impulse response vector. The first vector (i) is populated with sampled signal data of the received signal, r(n). The second vector (ii) is populated with signal data corresponding to the exponential function of the matched filter.
The vectors are initially populated with data. For the received signal vector, in Step 1 (
For the target impulse response vector, in Step 2 (
Next, in Step 3 (
The two vectors are now ready for comparison. The corresponding indices'"'"' value of two vectors are multiplied together in Step 4 (
Next, in Step 6 (
Steps 46 are repeated in Step 7 (
In Step 8 (
The processing can be repeated all over again for new vectors values in Step 9 (
The aforementioned convolution process can be executed using Matlab code using complex conjugate operations, for instance. The time reversal and complex conjugate operations are performed so that the filtering operation (i.e. convolution process) becomes a correlation between the exponential of the transmit signal, and the expected target response (i.e., harmonic signal).
Consider an example of isolating a desired second harmonic using a matched filter for a second power exponential function of transmit waveform function, for instance. For the second harmonic, the exponential power is 2. If ƒ(n) is the transmitted waveform, then ƒ^{2}(n) is the squared transmitted function. The square of the transmit waveform, ƒ^{2}(n), is mathematically defined as ƒ(n)×ƒ(n). The matched filter would thus be configured to look for the frequency of ƒ^{2}(n) which correspond to the second order (nonlinear) harmonic response in the received signal. The matched filter for the square of the transmit waveform function may be created as follows: ƒ_{m}(n) =m(n)*, where m(n)=ƒ^{2}(N−n), ƒ(n) is the transmitted waveform, m(n)* is the complex conjugate of m(n), N=2L, L is the length of ƒ(n), and the additional L time domain samples (not occupied by ƒ ƒ^{2}(n), n≤L) are equal to zero. The idea here is to zeropad (i.e., add zeros to the end of the timedomain signal) in such a way that there is no “wraparound” when the circular convolution is performed. That is why L is the length of the function (vector) f(n), and N=2L.
The processing should also be able to handle a situation in which multiple frequency tones are all transmitted simultaneously and in which they all must be added together thereby resulting in production of cross terms. The multitone, analytic (complex) signal would be represented as a sum of complex exponentials at the various frequencies. Hence, the product of e^{jat}*e^{jbt }would yield e^{j(a+b)t }(where and b are two different frequencies). The matched filtering would correspond to selection of the frequency bins corresponding to the desired intermodulation products.
In addition, the process can handle multiple frequency tones transmitted sequentially, such as in a chirp function or other function. In this case, since only one frequency will be transmitted at a time, then there would be no crossterms present. That is, if B_{i }is frequency band i, and ƒ_{i }is center frequency i, then ƒ_{2}=nƒ_{1}, and B_{2}=nB_{1}. (For a second harmonic, ƒ_{2}=2ƒ_{1}, and B_{2}=2B_{1}. It is noted that this is a reasonable approximation to the case where the nonlinearity can be represented by a quadratic (i.e. squaring the transmitted waveform). We then look at the squared terms in the nonlinear target response. These correspond to frequencymultiplied terms (in the frequency domain).
In addition, the process can handle other commonly encountered radar waveforms as transmit waveforms including, but not limited to: pseudorandom noise waveforms, polyphase code waveforms such as Frank codes of varying length, Barker codes of varying lengths, P1 codes, P2 codes, and pseudorandom codes of various lengths.
In further embodiments, the matched filtering may be used in conjunction with sidelobereduction techniques. A recursive sidelobe minimization (RSM) technique is one of said techniques which may be used.
In this processing, ƒ_{RSM}(n) denotes the function for the final high range resolution (HRR) signature produced by the iterative RSM processing. Since the RSM processing saves the minimum value for each HRR cell over all iterations, it must be first initialized. Because there will be many iterations, an arbitrary very high number may be used. Thus, in Step 1, the values of ƒ_{RSM}(n) are initially set to an very high number, such as 10^{30}, for all n. The initial values obtained for each of the HRR cell will be saved.
Next, in Step 2, the variables needed for iterative sidelode reduction are set. These include: R, the number of cells to randomly excise during recursive sidelobe minimization; P, the number of recursive sidelobe minimization iterations (it is essentially an iteration limit); and B, the band of frequencies considered for matched filtering. According to some embodiments, P may be in the vicinity of 3050 (based on our past experience), R may be somewhere around 30% of the total number of cells (i.e., the total number of frequency bins), and B can be the frequencies around the particular harmonic of interest (the bins noted earlier). If too many samples are excised, then there may be a risk of missing a strong target return. In this case, B might be selected to correspond to a wavelength (frequency) that is wellmatched to the target dimensions/size.
Some processing is preferably performed in the frequency domain, instead of the time domain, which is made much simpler in terms of processing. It is noted that this same processing could be done in the time domain by storing a sequence of “filtered” match filters corresponding to the response obtained when a particular set of frequency bins has been removed. This would require saving a large number of coefficients, since a separate filter would be required for each set of randomly excised frequency bins.
Thus, in Step 3, the measured received signal, r(n), and the matched filter, ƒ_{m}(n), are both transformed into the frequency domain. This can be performed using a wellknown Fourier transform. The transformed signals are then denoted as R(m) and F_{m}(m), respectively. (As mentioned above, functions denoted in capital letters are often used for frequency domain versions of functions, i.e., Fourier transforms).
Sidelobe reduction is an iterative process. Thus, it will be repeated a number of times; previously P was define in the earlier step. This looping actually represents the RSM processing used. Each time through the loop, different samples are excised. This causes the sidelobe levels within the hrr signature to change while the target levels remain about the same. Hence, by keeping saving the minimum value over all iterations you end up with a target level that is about the same while the sidelobe/noise levels are reduced.
RSM is now ready to begin. The recursive or iterative steps are in collectively performed in Step 4. Variable P, the iteration limit, denotes the number of times these steps are repeated.
It is important to get different/diverse sets of excised samples so that the sidelobe patterns can change significantly from iteration to iteration. So a random sampling is taken each iteration. In step, Step 5, R frequencies samples are randomly excided from within frequency band B. There are functions in Matlab, rand( ) and randperm( ) for randomly selecting numbers within a given range of values. The random number generator can be used to select the samples (index numbers) that are to be excised (set to zero).
The presence of zero values in the frequency domain representation of a filtered signal induces a particular sidelobe pattern in its time domain representation. This corresponds to its HRR profile. F_{out}(m) will denote the frequencydomain representation of the HRR profile ƒ_{RSM}(n) produced by each iteration of the RSM processing. Thus, F_{out}(m) will be zero in the randomly selected bins described above.
In Step 6, sidelobe reduction is applied to F_{out}(m). Sidelobe reduction may be thought of as being a smoothing process of sorts. In this case, a windowing operation can be used in the processing here for reducing sidelobe levels (at the expense of an increased mainlobe width). In general, it includes multiplying the function by another function that is tapered at the ends (like a bellshaped curve). The idea is to avoid abrupt transitions (to zero) at the ends of the data sequence. Example windowing functions W(m) can include the Matlab functions: hanning( ) for a Hanning window, taylorwin( ) for a Taylor window, and rectwin( ) for a rectangular window.
Apply frequency windowing and calculate F_{out}(m)=W(m)R(m) F_{m}(m), where W(m) is a suitable weighting window function. R(m)*F_{m}(m) in this calculation is actually the matched filtering convolution calculated in the Frequency domain.
Ranging distances are computed from the previouslyformed F_{out}(m) in Step 7. Intensities for various ranging distances are represented here as the complex magnitude of ƒ_{out}(n), ƒ_{out}(n). As previously stated, F_{out}(m) is in the frequency domain. Thus, to covert the frequency domain to the ranging domain, a wellknown inverse fast Fourier transform (IFFT) may be used here. Matlab provides function: IFFT (X), which returns the inverse discrete Fourier transform of vector X, computed with a fast Fourier transform (FFT) algorithm. So ƒ_{out}(n)=IFFT(F_{out}(m)).
The RSM processing saves the minimum value for each HRR cell across all iterations. As such, the previously stored ƒ_{RSM}(n) value and current value of ƒ_{out}(n) for this iteration are compared for each cell and the smaller selected in Step 8. For every n, ƒ_{RSM}(n)=min(ƒ_{RSM}(n), ƒ_{out}(n)). Put simply, this means that we take the smallest of thesavedoutput magnitude from the matched filter and current output magnitude of the matched filter, recalling that the saved output magnitude represents the minimum value across all previous iterations.
The final result of the processing, in Step 9, is that the high resolution range profile (with reduced sidelobe levels), ƒ_{RSM}(n) as output by the system. ƒ_{RSM}(n) is real.
For the aforementioned embodiment, RSM has been integrated directly into the high resolution range (HRR) profile generator. This modified processing improves the localization capabilities of the resulting system. In addition, the suppression of sidelobe artifacts also makes target detection more reliable by enhancing the contrast between target and nontarget elements of the HRR profile. This is illustrated in
In
The novel methodology eliminates natural clutter by processing nonlinear responses. Many of the potential false alarms plaguing modern radar systems can be discarded in the earliest processing stages. That is, natural clutter responses would be virtually eliminated from the processed radar data sets. Natural clutter is responsible for many false alarms. Many radar, target detection algorithms employ multiple stages. So, the false alarms attributable to natural clutter would be eliminated by the earliest algorithms, e.g., a prescreener of some sort. In addition, the system exploits an adaptive, matched filtering formulation combined with a recursive sidelobe minimization strategy. This processing paradigm enables both localization of the target and enhancement of the local targettobackground ratio. These, in turn, serve to increase target detectability.
The system processing increases targettoclutter ratios by exploiting the nonlinear characteristics of specific target responses to a suitable radar probe signal. Qualitatively, the methodology is straightforward: natural clutter does not produce a nonlinear response; hence, there would be (in theory) an infinite targettoclutter ratio for nonlinear targets. It also leverages radar processing techniques previously configured for use with linear systems. These techniques provide additional contrast between the target response and responses due to background clutter and artifacts (e.g. sidelobes). Existing linear systems are, by their very nature, unable to exploit the nonlinear target responses. Hence, they cannot realize the increases in targettoclutter ratio achievable in the nonlinear system.
The novel methodology may be used with any application requiring location of electronic devices within an area under surveillance. This could include an initial scan of the area to ensure no devices are present. It could also include standoff interrogation of people/materiel entering a restricted area. Of course, the target of interest could include isolated areas in which the presence of electronics could indicate danger.
It is noted that aspects relating to this invention have been previously disclosed in:
(i) Gallagher, K. A.; Mazzaro, G. J.; Ranney, K. I; Nguyen, L. H; Martone, A. F.; Sherbondy, K. D.; and Ram M. Narayanan, R. M., “Nonlinear synthetic aperture radar imaging using a harmonic radar,” Proc. SPIE 9461, Radar Sensor Technology XIX; and Active and Passive Signatures VI, 946109 (May 21, 2015); doi:10.1117/12.2177219, publishing the conference proceedings of Apr. 20, 2015; and
(ii) Gallagher, K. A.; Narayanan, R. M.; Mazzaro, G. J.; Ranney, K. I.; Martone, A. F.; and Sherbondy, K. D., “Moving target indication with nonlinear radar,” presented at the Radar Conference (RadarCon), 2015 IEEE, of 1015 May 2015, the disclosures of which are herein incorporated by reference in their entities.
The foregoing description, for purpose of explanation, has been described with reference to specific embodiments. However, the illustrative discussions above are not intended to be exhaustive or to limit the invention to the precise forms disclosed. Many modifications and variations are possible in view of the above teachings. The embodiments were chosen and described in order to best explain the principles of the present disclosure and its practical applications, to thereby enable others skilled in the art to best utilize the invention and various embodiments with various modifications as may be suited to the particular use contemplated.
Various elements, devices, modules and circuits are described above in associated with their respective functions. These elements, devices, modules and circuits are considered means for performing their respective functions as described herein.
While the foregoing is directed to embodiments of the present invention, other and further embodiments of the invention may be devised without departing from the basic scope thereof, and the scope thereof is determined by the claims that follow. All references cited above are hereby incorporated by reference herein for all purposes.