US007 199750B2

(12) United States Patent (10) Patent No.: US 7,199,750 B2

Bourdelais et al. (45) Date of Patent: Apr. 3, 2007

(54) REAL-TIME MULTISTATIC RADAR SIGNAL 3.996,590 A * 12/1976 Hammack ................... 342/.465
PROCESSING SYSTEMAND METHOD 4,499.468 A 2f1985 Montana et al.
5,097,269 A * 3/1992 Takayama et al. .......... 342/453
(75) Inventors: John Michael Bourdelais, Vienna, VA 5,247,307 A 9, 1993 Gandar et al.
(US); Ernest Scott Stickels, Lothian, 6,218,979 B1 4/2001 Barnes et al.
MD (US); William Ray Wright 6,373,426 B1 4/2002 Hellsten
Springfield VA (US); David E arl 6,614,384 B2 9/2003 Hall et al.
Norris, Fairfax Station, VA (US); 6,661,342 B2 12/2003 Hall et al.
Michael Anthony Tiberio, Arlington, 2003/0006930 A1* 1/2003 Lodwig et al. ............. 342,120
2003/0085840 A1* 5/2003 Benner et al. .... ... 342/450
VA (US); Gary Dana Butler, 2004/0233105 A1* 11/2004 Benner et al. .... ... 342/451
Alexandria, VA (US) 2005, OO12657 A1* 1/2005 Mohan .......... ... 342,133
2006/0227042 A1* 10, 2006 Budic ......................... 342,195
(73) Assignee: BBN Technologies Corp., Cambridge,
MA (US) FOREIGN PATENT DOCUMENTS
(*) Notice: Subject to any disclaimer, the term of this DE 3237 137 9, 1983
patent is extended or adjusted under 35
U.S.C. 154(b) by 104 days. (Continued)
(21) Appl. No.: 11/112,390 OTHER PUBLICATIONS
(22) Filed: Apr. 22, 2005 "Netted radar and the ambiguity function', Papoutsis, I., Baker,
C.J.; Griffiths, H.D. Radar Conference, 2005 IEEE International
(65) Prior Publication Data May 9-12, 2006 pp. 883-888.*
US 2006/0238407 A1 Oct. 26, 2006 (Continued)
Primary Examiner John B. Sotomayor
(51) Int.
GOISCl.I.3/08 (2006.01) (74) Attorney, Agent, or Firm Proskauer Rose LLP
(52) U.S. Cl. ............................ 342/90; 342/95; 342/96; (57) ABSTRACT
342/97; 342/108; 342/146; 342/126; 342/140;
342/107; 342/133: 342/189; 342/195 A real-time signal processing engine robustly detects, local
(58) Field of Classification Search ............ 342/90–99, izes, tracks and classifies ground targets based on radar
342/59, 108, 146, 126, 147, 140, 158, 107, signals from a multistatic radar system. The system differ
342/135, 131, 133, 189, 195 entiates between different targets based on an optimized cost
See application file for complete search history. function, which can include the total returned normalized
(56) References Cited pulse energy. The local transmitters/receivers can commu
U.S. PATENT DOCUMENTS
nicate with each other via the transmitted radar signals.
3.242,487 A 3, 1966 Hammack ................... 342,103 21 Claims, 6 Drawing Sheets

Start
502

Refer
f strial from a
caster K
strine islatanga
54
- datamine line-of-sight
4 selectellipsecision
within
crossings
window
- -518
distance transite-receivers an as
deterris ifiers detected
1 : - - 520
selectrose crossing -
signals and scalaisdaptively having the greatest total ;
filterisubtracts previous normalizid signal 8F8th
referencescar
508 -------------------------------------
522
elermine potentiattack
based offiHercritair
wi
crossir

510 X. 524
Remove tipses associated
Repeat with the most likely ellipse
featkast two
crossing front sat of
raisfitter-feelw8
determined slipses
pairs
512
All
ellipse ceasings
pfixessed
dearmine elipses
associatadw:h the is from
the atlasst tw83tter
facsiwer pats Yes 528

50
- Create targetinap and
refine with
 US 7,199,750 B2
Page 2

FOREIGN PATENT DOCUMENTS Mark G. Rutten, et al., “Multipath Track Association for Over-the
Horizon Radar using Lagrangian Relaxation'. Commonwealth of
GB 2251149. A * 6, 1992 Australia. (2004).
JP O1217285 A * 8, 1989 PulsONTM Technology, “Time Modulated Ultra-Wideband For
JP 2003248056 A * 9, 2003 Wireless Applications'. Time Domain Corporation, May (2000).
WO 2002/23218 3, 2002
Herman et al., “A Particle Filtering Approach to FM-band Passive
OTHER PUBLICATIONS Radar Tracking and Automatic Target Recognition.” Aerospace
Conference Proceedings, IEEE Mar. 2002.
Jeremy S. DeBonet, et al., “MIMIC: Finding Optima by Estimating Invitation to Pay Additional Fees.
Probability Densities' Advances in Neural Information Processing Notification of Transmittal of The International Search Report and
Systems, 1997, MIT Press, Cambridge. The Written Opinion of the International Searching Authority, or the
Robert J. Fontana, “Recent Applications of Ultra Wideband Radar Declaration.
and Communications Systems”, Kluwer Academic/Plenum Publish
S. * cited by examiner
U.S. Patent Apr. 3, 2007 Sheet 1 of 6 US 7,199,750 B2

-- saySV 9
A-Z ZZZZZ

has
W saw V 92
47/AAA

Nessy V 9
- ZZAAAA in
U.S. Patent Apr. 3, 2007 Sheet 2 of 6 US 7,199,750 B2
U.S. Patent Apr. 3, 2007 Sheet 3 of 6 US 7,199,750 B2
U.S. Patent Apr. 3, 2007 Sheet 4 of 6 US 7,199,750 B2
U.S. Patent Apr. 3, 2007 Sheet S of 6 US 7,199,750 B2
U.S. Patent Apr. 3, 2007 Sheet 6 of 6 US 7,199,750 B2

1 514
Receiving at a receiver a
radar signal from a
- 516
corresponding transmitter Determine bistatic angle

518
Select ellipse Crossings
Determine line-of-sight within an association window :
distance transmitter-receiver init

Determine t for a detected

ill- 520
Select ellipse crossing
signals and scalefadaptively having the greatest total
filterisubtract a previous normalized signal strength
reference scan
522
Determine potential track Associate target location with
based on filter Criterion the most likely ellipse
crossing

524
Remove ellipses associated
Repeat with the most likely ellipse
for at least two crossing from set of
transmitter-receiver determined ellipses
pairs

AI
ellipse crossings
processed
Determine ellipses
associated with the 's from
the at least two transmitter
receiver pairs YES 528

so Create target map and

refine with MHT

FIG. 6
 US 7,199,750 B2
1. 2
REAL-TIME MULTISTATIC RADAR SIGNAL According to another aspect, the invention provides a
PROCESSING SYSTEMAND METHOD method for identifying a target using a multistatic radar
system having at least one radar transmitter and a plurality
STATEMENT REGARDING FEDERALLY of radar receivers forming node pairs. The method includes
SPONSORED RESEARCH AND receiving target data from a target in at least two receivers
DEVELOPMENT defining node pairs, wherein the received target data define
a target ellipse at each of the node pairs, by determining
This invention was made with government Support under points of intersection between the target ellipses, associating
DARPA Contract Number MDA972-03-C-0083. The Gov a cost function with the points of intersection, and assigning
ernment has certain rights in the invention. 10 a first target to the point of intersection having the lowest
cost function.
BACKGROUND OF THE INVENTION According to a further aspect, the invention provides a
method for classifying a target using a multistatic radar
The invention is directed to a multistatic radar system system having at least one radar transmitter and a plurality
with real-time signal processing for detecting, localizing, 15 of radar receivers arranged in node pairs. The method
characterizing and tracking ground targets of interest. includes receiving at at least two receivers radar signals
Ground clutter and signals from stationary targets are iden from the transmitter of a corresponding node pair, associ
tified and removed at the local radar receivers. ating the received radar signals with a target and with a
In many situations, there is a need to monitor ground relative motion of parts of the target, determining a location
based objects and targets, for example, to prevent home of the target, and classifying the target based on the radar
intrusion, burglaries, and for military applications, such as signals received from a target and the relative motion of the
perimeter surveillance. Often, it is important to not only parts of the target.
detect these targets, but also to discriminate between targets. According to another aspect, the invention provides a
Targets of interest can be, for example, humans on foot, multistatic radar tracking system with at least one radar
wheeled or tracked vehicles. Such surveillance poses par 25 transmitter and a plurality of radar receivers arranged in
ticular challenges in foliage environments as radio fre node pairs. The transmitter emits radar signals composed of
quency (RF) energy experiences attenuation, dispersion, and ultra-short wide bandwidth radio frequency pulses. The
modulation, making it difficult to detect and/or discriminate system further includes a filter for identifying at the receiver
targets from Stationary and non-stationary clutter. Single level from a radar signal received at the receiver of a node
narrowband monostatic radar systems tend to have poor 30 pair a target ellipse for a potential target. The system also
range resolution, and in spite of higher transmitted power includes a real-time radar signal processing engine for
have limited coverage area in foliage environments due to determining from the at least two target ellipses a first target
attenuation and multi-path fading. location, and for determining an additional target location
Accordingly, there is a need for a radar tracking system after removing the target ellipses used to determine the first
that can efficiently detect and track moving targets, in 35 target location.
particular multiple targets, and discriminate between targets Embodiments of the invention may include one or more
and stationary and moving clutter under adverse environ of the following features. The received radar signals may be
mental conditions. There is also a need to distinguish filtered to Suppress out-of-band energy, to time-align Suc
between and classify different types of targets, for example, cessively received radar signals and/or to Scale, Smooth and
humans, animals, and vehicles of different size. 40 normalize the received signals. Probable target data may be
identified by determining backscattered energy of the
SUMMARY OF THE INVENTION received radar signal and by comparing the backscattered
energy with a predetermined reference value. Probable tar
The invention addresses the deficiencies in the prior art get data may also be identified by determining a target speed
by, in various embodiments, providing methods and systems 45 and comparing the determined target speed with a probable
for detecting, localizing, identifying, characterizing, and/or target speed of the target.
tracking one or more ground targets by real-time processing The location of the first target may be identified by
of ultra-wideband signals from a multistatic radar system. determining from the probable target data a total backscat
The forward-scattered or backscattered radar signals are tered energy, and selecting as the location of the first target
filtered and processed at the node level to eliminate signals 50 the location having a greatest total backscattered energy.
unlikely to be associated with an actual target. A target Alternatively or in addition, the location of the first target
ellipse for a potential target is also identified at the node may be identified by associating a cost function with poten
level from the received radar signals. tial target locations, wherein the cost function represents a
According to one aspect, the invention provides a method measure of likelihood that a target is present at the location,
for identifying a target using a multistatic radar system 55 and by selecting the location of the first target by optimizing
having at least one radar transmitter and a plurality of radar the cost function. For detecting multiple targets, the target
receivers arranged in node pairs. The method includes ellipses that include the target data for the first target may be
receiving at a receiver a radar signal from a corresponding removed and a second target may be assigned to the point of
transmitter of the node pair, filtering the received radar intersection having the next-lowest cost function. An exem
signal and identifying from the filtered radar signal probable 60 plary cost function may include a received normalized pulse
target data, transmitting the probable target data from at least energy, a bistatic angle for the received pulse energy, a
two transmitter-receiver node pairs to a central node, iden range—velocity characteristic of the target, and/or a sepa
tifying at the central node from the probable target data a ration between closely-spaced points of intersection located
location of a first target, removing from the probable target on at least three intersecting ellipses.
data the data associated with the first target location, and 65 Previously known or previously measured forward-scat
determining a location of an additional target from the tering and backscattering radar cross sections may be used
remaining probable target data. to estimate a range for a likely bistatic angle for the probable
 US 7,199,750 B2
3 4
target data, which may narrow the parameter space to be enabling a more precise target localization than monostatic
considered and may also disambiguate target locations. radar systems, combined with system robustness in the event
A target can be classified by tracking an identified target of a node failure, i.e., the failure of one transmitter-receiver
over a predetermined time, building a three-dimensional node pair. The central base station 18 can control the various
image of the target by analyzing target data received from a transmitters/receivers 12, 14, and 16 and perform central
plurality of the node pairs, monitoring changes in the data collection and data processing, also referred to as
three-dimensional image during the predetermined time, and “field-level control.” Any one of the illustrative transmitters/
assigning a target type to the target based on the monitored receivers 12, 14, and 16 may be configured to operate as
changes. Potential target types may be, for example, receiver or transmitter or both, and may include a respective
vehicles, livestock, and humans, which can be differentiated, 10 filter 22, 24, 26 for processing at the receiver level a forward
for example, based on their size and radar cross section. In or backscattered radar signal associated with a potential
addition, the node pairs may be configured for receiving target.
radar signals from different vantage points to form a three Referring now to FIG. 2, an illustrative multistatic UWB
dimensional radar image of the target. Further, the target can radar system 11 includes at least one UWB transmitter 100
be viewed from multiple aspects (using one or more receiv 15 and several UWB receivers 110, 120, 130, whereby any two
ers) and its radar cross section monitored for changing levels of the transmitter/receiver node pairs {100/110}:{100/120):
enabling classification. and {100/130} function as a bistatic radar system to facili
Further features and advantages of the invention will be tate target detection. The introduction of an object, Such as
apparent from the following description of illustrative an intruder 102, can alter the scan waveform received by one
embodiments and from the claims. or more of the receivers 110, 120, 130 compared to the scan
waveform without the intruder 102. The presence of the
BRIEF DESCRIPTION OF THE DRAWINGS intruder 102 is now detectable in the subtraction of succes
sive scans; any significant change in a portion of this
The following figures depict illustrative embodiments of difference reveals the range of the intruder 102 with respect
the invention in which like reference numerals refer to like 25 to the placement of a scanning receiver, such as receiver 110.
elements. These depicted embodiments are to be understood Knowing the direct distance between the transmitter 100 and
as illustrative of the invention and not as limiting in any way. the scanning receiver 110, which can be determined from the
The use of the same reference symbols in different drawings time-of-flight of the direct signal path 104 between trans
indicates similar or identical items. The drawings may not be mitter 100 and receiver 110, and knowing the relative time
to Scale. 30 delay of the target response in the Scanned waveform (which
FIG. 1 shows Schematically an exemplary configuration travels from the transmitter 100 to the target 102 along path
of a multistatic radar system; 106, and from the target 102 to the receiver 110 along path
FIG. 2 shows a basic diagram of possible target ellipses 108) with respect to the direct signal path 104, the position
and target localization with a multistatic radar system; of the intruder 102 is known to lie somewhere on an ellipse
FIG. 3 shows a basic diagram when multiple targets are 35 111 whose foci are the transmitter 100 and the receiver 110.
present; Likewise, the signal returned from target 102 travels
FIG. 4 shows a basic diagram after elimination of a first along path 118 to receiver 120, and along path 128 to
target with the method of the invention; receiver 130, with the direct distance 114 between transmit
FIG. 5 shows a basic diagram for differentiating between ter 100 and receiver 120, and the direct distance 124
targets based on backscattered and forward Scattered energy 40 between transmitter 100 and receiver 130 determined as
with the method of the invention; and before. The target coordinates of target 102 can thus be
FIG. 6 is a process flow diagram for determining multiple unambiguously determined from the point of intersection
targets with the method of the invention. between the three ellipses 111, 121, and 131. It is also
evident in FIG. 2 that signals received by only two trans
DESCRIPTION OF ILLUSTRATED 45 mitter-receiver node pairs may result in ellipses that inter
EMBODIMENTS sect at more than one point or potential target location,
causing ambiguities unless further information is utilized.
As described above in Summary, the invention provides, Techniques used to resolve this ambiguity are discussed
in various embodiments, systems and methods for detecting below.
and tracking one or more moving targets using multistatic 50 The illustrative systems and methods include a radar
ultra-wideband radar, and for distinguishing between and/or signal processing engine (RSPE) capable of detecting targets
classifying different types of targets. of interest and filtering out clutter. Data is generated from a
FIG. 1 shows schematically a multistatic radar system 10 radar scan by measuring the backscattered energy as a
having one or more transmitters, for example, transmitter function of range at a predetermined scan rate. For certain
12, emitting ultra-short wide bandwidth radio frequency 55 transmitter-receiver-target geometries where the target is
pulses and a plurality of distributed receivers, for example, near the line connecting the transmitter to the receiver, the
receivers 14, 16 capable of detecting forward scattered forward Scattered energy is incident on the receiver, typi
and/or backscattered radar signals representing the reflec cally resulting in an enhanced signal level. Scan data col
tions from the local environment. Further, each transmitter lected by a particular transmitter-receiver node pair is first
and/or receiver can be configured to communicate radar 60 filtered with a bandpass filter or matched filter to suppress
signal information with another transmitter and/or receiver out-of-band energy. The scans are then time-aligned relative
and/or with a central base station 18. The communication to the direct arrival with reference scans (without a detected
between the transmitters/receivers can be point-to-point or target). Signals can be adaptively scaled, for example based
via a network 20, and can be wired or wireless, using, for on an exponential Smoothing filter combination of previous
example, conventional communication protocols, such TCP/ 65 scans, to minimize false residuals after Subtraction from the
IP. Distributed receivers can provide diverse views of a current Scan. The magnitude output after Subtraction is then
target at any one location from different observation points, normalized by the rms level in each transmitter-receiver cell
 US 7,199,750 B2
5 6
to obtain a constant false alarm rate (CFAR). The subtraction locations, except for the true target location 402a, the ellipse
process removes stationary clutter; however, moving clutter, association logic in the field processor can disambiguate
Such as tree branches, can still persist through the Subtrac between the localization points 402a, 402b, 402c, and 402d
tion and potentially create unwanted false alarms. To elimi based on the radar cross section from all ellipses, as
nate false alarms or at least reduce the frequency of false described above with reference to FIGS. 3 and 4.
alarms, any potentially detected target with a returned signal The target ellipses can be removed at the local (transmit
energy exceeding a predetermined threshold value passes ter/receiver pair) level by disregarding, for example, pulses
through a range delay tracker that determines if over time the returned within the time window represented by the
potential target has features (such as a predictable ground removed ellipses. This can advantageously be achieved by
speed of certain types of moving targets, e.g. vehicles, 10 embedding communications in the radar transmission, thus
intruders on foot) characteristic of a genuine target. Only obviating the need for separate communication channels
those targets that meet the tracker filter criterion are passed between transmitters and receivers. The communications
on to a field level processor as potentially genuine targets. allow for information about node pair tracks to be shared so
Targets exhibiting changes in location that are either unex that localization can be done at the field level, with the
pectedly slow or unexpectedly fast are dismissed. 15 information to be sent to an end user.
FIG. 3 shows a situation where the illustrative multistatic In the determination of the most likely target, as in many
UWB radar system 21 detects several targets, such as the optimization problems, the structure of potential Solutions
two exemplary targets 102, 202, which meet the tracker filter can reflect complex relationships between the different input
criterion at each receiver. The range ellipses associated with parameters. While some parameter values may be judged to
each target and the total normalized pulse energy returned independently represent valid input data, other parameters
from the targets 102 and 202 and measured at receivers 110. may be closely related and should not be explored indepen
120, 130 are passed to a field level processor. The field level dently. It is therefore customary to assign to complex
processor associates the range ellipse intersections which are Solutions a cost function that has local minima dependent on
within a preset distance of one another with a potential the input parameters, with the goal to find those input
target. A single ellipse may be associated with several 25 parameter values that optimize the cost function. Input
potential targets (multiple target hypothesis). The energy parameters contributing to the cost function may be, for
passed from each receiver for the multiple range ellipses are example, the tracker filter criterion, the received normalized
Summed for each potential target and used as a measure for pulse energy, the expected bistatic angle for the received
the likelihood of having detected an actual target. For pulse energy, the spread or diameter of the circle defined by
example, it will be assumed that the returned normalized 30 intersecting ellipses, and the like.
pulse energy from target 102 is greater than from target 202. An efficient algorithm for detecting and discrimination
Accordingly, target 102 will be assumed to be a valid target among different moving targets will be described below.
and will be indicated as such on a target map. Thereafter, the Referring now to FIG. 6, a process 50 for detecting
target ellipses 111, 121, 131 are removed from the set of multiple moving targets using UWB multistatic radar having
measured ellipses, leaving target ellipses 211, 221, 231 that 35 at least one transmitter and a plurality of receivers includes
define target 202, as indicated in FIG. 4. The process then receiving at two or more receivers (as mentioned above, a
repeats, with target 202 now being the target generating the minimum of three receivers is required to resolve ambigu
greatest normalized pulse energy of the remaining targets ities) an ultra-short, i.e., wide bandwidth radio frequency
(not shown). radar pulse that can be encoded for communications
It may arise that the three ellipses created by, for example, 40 between transmitter and receivers, step 502. With properly
target 102 in FIG. 2, come within a preset distance from each synchronized timing between transmitter(s) and receivers, a
other, as indicated in FIG. 2 by the circle 103, thus creating direct path signal received by a receiver from the transmitter
an ambiguity in the target definition. Two techniques can be defines a line-of-sight distance between the transmitter and
used to address this issue. The first technique determines, for the associated receiver, step 504. The emitted radar pulses
example, which prospective target location defined by the 45 are backscattered or forward scattered from stationary and
multiple ellipses has the lower spread, i.e., a smaller circle moving objects, with the time difference At between the
103, in the intersection. scattered signal and the direct signal defining an ellipse
The second technique depicted in FIG. 5 takes advantage having the corresponding transmitter and receiver as foci,
of the increase in signal strength from a target having a high step 506. The scattered radar data are processed in a matched
bistatic angle. As shown in FIG. 5, an exemplary target 402a 50 filter using a complex finite-impulse-response (FIR) filter
with a high bistatic angle C. is positioned near the line that uses the averaged time series response of the direct path
connecting the transmitter 400 and the receiver 410. Another arrival as filter weights. The weights are convolved with a
potential target 404 is shown as having a low bistatic angle complex bandpass filter to further reduce band noise.
C'. The radar cross section (RCS) of each target is estimated Further in step 506, an adaptive filtering/subtraction algo
using range and location information. The target 402a with 55 rithm is employed to generate a reference waveform that
the high bistatic angle a located on the range ellipse 40 represents consistent, stationary energy that is Subsequently
produces an enhanced received signal level which is asso subtracted from scans collected by the receiver. The refer
ciated with the four equivalent positions 402a, 402b, 402c, ence waveform can be constructed, for example, from a
and 402d. If the signal enhancement as a function of bistatic predetermined number of preceding scans and will therefore
angle is known, then the ambiguity between the low bistatic 60 Substantially only show stationary backscattering objects. A
angle O' target location 404 (there are three additional range-dependent exponential Smoothing algorithm with a
ambiguous low bistatic angle positions, which have been convergence rate constant that decreases with range is used.
omitted in FIG. 4) and the high bistatic angle C. locations Exponential Smoothing is a simple algorithm for predicting
402a, 402b, 402c, and 402d can be reduced from eight to the next value in a time series given the current value and the
four, namely to locations 402a, 402b, 402c, and 402d. 65 current prediction.
Since the four ambiguous points 402a, 402b, 402c, and Each range cell (i.e., an area Surveyed) has a certain noise
402d for each source-receiver ellipse will be at different level which can increase for cells near the direct signal path,
 US 7,199,750 B2
7 8
for example, due to residual energy from the Subtraction guities between, for example, signals arising from the same
process described above. Therefore, each cell is advanta target and traveling via separate propagation paths, and
geously normalized by a long-time average power in the signals arising from different targets.
cell, with the normalization also aiding in maintaining a Several techniques can be used to classify an identified
constant false alarm rate. target. One technique is to build a three-dimensional image
The disclosed systems and methods are optimized for of a target, as it is tracked over time, from the multiple
operating in dense foliage which can result in unwanted vantage points provided by the plurality of receivers. For
multi-path returns and extended regions of high levels of example, characteristic parameters of expected targets cor
residual clutter. To this end, a split window normalization relating their radar cross section with their size and location
routine can be applied as a filter criterion in step 506, which 10 can be used. The parameters can be acquired, for example,
operates for each scan across the range delay window of in field tests and stored at the node level. Additionally, the
interest and estimates a normalization factor for each range relative motion of parts of targets can be analyzed and
delay cell. On a time axis, the energy in a window to the left potentially differentiated at the node level by examining the
(shorter times) and in another window to the right (longer received signal energy over multiple target data acquisitions.
times) of the range delay window is measured. The width of 15 For example, for determining movement of a part of a target,
the left and right window is selected so as to measure the Such as a limb of a person, relative to the rest of the target,
average power without introducing errors due to fluctuations i.e. the person's body, a predetermined number of scans
in background power. If the right window has significantly preceding the current scan may be time-shifted to detect the
more energy than the left window (indicative of significant change in location of the limb over time and thereby its
multipath returns), then only the left window is used; Velocity, where after the energy in the ranges of interest are
otherwise the average from the two windows is used as a summed. For example, a human would show different
noise normalization factor. In this way, target returns that Velocities, when Swinging his arms or legs, which would be
have Substantial multipath returns following the initial peaks different from those of, for example, a deer. These signals
are preserved. could be distinguished from those produced, for example, by
Target tracking may be done in three stages. In a first stage 25 a small vehicle where all the parts would move in unison.
on the node level, a transmitter-receiver node pair generates Moreover, the size of different targets and the material from
radar scan data vs. time that is processed according to the which they are made can cause different return signal
routines outlined in steps 506 and 508. The node level strengths for a given range and aspect angle, further aiding
tracking algorithm associates successive peaks that meet in the differentiation between, for example, animals and
certain criteria, Such as target speed, with the initially 30 vehicles. Further, targets viewed from different aspects may
detected target. Tracks having normalized signal levels present differing radar cross sections depending on the
below the predetermined threshold value are eliminated. vantage point. For example, a human target on foot will have
The range delay At of each transmitter-receiver node pair a relatively constant RCS when viewed from different angles
generates an ellipse of possible locations, step 512. Each while a vehicle will exhibit a varying RCS depending upon
possible combination of ellipses determined in step 512 is 35 the angle from which it is viewed.
examined for each target detection to determine the most While the invention has been disclosed in connection with
likely target location, for example, by optimizing a cost the preferred embodiments shown and described in detail,
function, step 514, indicated by the dashed box. The cost various modifications and improvements may be made
function may have multiple input variables representing a thereto without departing from the spirit and scope of the
space of selection criteria, with the most likely target loca 40 invention. By way of example, although the illustrative
tion and/or target class or type obtained through optimiza embodiments are depicted as having a fixed number of
tion of the cost function. Determining the cost function may transmitters and/or receivers, this need not be the case.
include determining the bistatic angle of the target based on Instead additional UWB radar transmitters and/or receivers
the target signal level and radar cross section, step 516, may be integrated in the network, whereby each transmitter/
selecting the combination of ellipses producing a crossing 45 receiver may advantageously be configured with its own IP
within an association window limit, step 518, and/or iden address. The various transmitters and/or receivers may be
tifying the ellipse crossing with the highest total returned connected by wires, optical fibers, and/or wireless. Accord
normalized signal energy, step 520. The signal energy may ingly, the spirit and scope of the invention is to be limited
be normalized to produce a signal indicative of for example, only by the following claims.
the size of a target. The exemplary steps 516, 518, and 520 50
What is claimed is:
in FIG. 5 may be executed in any suitable order, and other
suitable selection criteria may be added. 1. A method for identifying a target by using a multistatic
Once the ellipse crossings are associated with a particular radar system having at least one radar transmitter and a
target location, Step 522, the ellipses for that target location plurality of radar receivers arranged in node pairs, the
are removed from the set of detected ellipses and are no 55 method comprising:
longer considered for additional target localizations, step receiving at a receiver a radar signal from a corresponding
524. If all potential ellipse crossings have been examined, transmitter of the node pair;
step 526, then the process terminates in step 528, optionally filtering the received radar signal and identifying from the
producing a target map of target positions that may be filtered radar signal probable target data;
transmitted for display to another location, for example, to 60 transmitting the probable target data from at least two
a central observation room. Otherwise, the process 50 transmitter-receiver node pairs to a central node:
returns to step 514 to optimize the cost function for the identifying at the central node from the probable target
remaining ellipse crossings for determining another poten data a location of a first target;
tial target. removing from the probable target data the data associ
A multi-hypothesis tracker (MHT), indicated in step 528, 65 ated with the first target location; and
can be used to reduce false alarms and refine target track determining a location of an additional target from the
estimation. Multi-hypothesis techniques can resolve ambi remaining probable target data.
 US 7,199,750 B2
9 10
2. The method of claim 1, wherein filtering comprises determining points of intersection between the target
Suppressing out-of-band energy of the received signal. ellipses;
3. The method of claim 1, wherein filtering comprises associating a cost function with the points of intersection;
time-aligning Successively received signals. and
4. The method of claim 1, wherein filtering comprises 5 assigning a first target to the point of intersection having
Scaling, Smoothing and normalizing the received signals. the lowest cost function.
5. The method of claim 1, wherein identifying probable
target data comprises determining backscattered energy of 16. The method of claim 15, wherein the cost function
the received signal and comparing the backscattered energy comprises at least one of a received normalized pulse
with a predetermined reference value. 10 energy, a bistatic angle for the received pulse energy, a
6. The method of claim 1, wherein identifying probable range—velocity characteristic of the target, and a separation
target data comprises determining a target speed and com between closely-spaced points of intersection located on at
paring the determined target speed with a probable target least three intersecting ellipses.
speed of the target. 17. The method of claim 15, further comprising
7. The method of claim 1, wherein identifying the location 15 removing the target ellipses that include the target data for
of the first target comprises determining from the probable
target data a total backscattered energy, and selecting as the the first target; and
location of the first target the location having a greatest total assigning a second target to the point of intersection
backscattered energy. having the lowest cost function.
8. The method of claim 1, wherein identifying the location 20 18. The method of claim 15, wherein the node pairs
of the first target comprises associating a cost function with receive the target data from different vantage points, said
potential target locations, said cost function representing a target data indicative of a size and a radar cross section of
measure of likelihood that a target is present at the location, a target.
and selecting the location of the first target by optimizing the
cost function. 25 19. A method for classifying a target by using a multistatic
9. The method of claim 1, comprising classifying a target radar system having at least one radar transmitter and a
by plurality of radar receivers arranged in node pairs, the
tracking an identified target over a predetermined time; method comprising:
building a three-dimensional image of the target by ana receiving at at least two receivers radar signals from the
lyzing target data received from a plurality of the node 30 transmitter of a corresponding node pair;
pairs; associating the received radar signals with a target and
monitoring changes in the three-dimensional image dur with a relative motion of parts of targets;
ing the predetermined time; and
assigning a target type to the target based on the moni determining a location of the target; and
tored changes. 35 classifying the target based on the radar signals received
10. The method of claim 9, wherein the assigned target from a target and the relative motion of the parts of
type is selected from the group consisting of vehicles, targets.
livestock, and humans. 20. The method of claim 19, wherein the node pairs
11. The method of claim 9, wherein the target type is receive the radar signals from different vantage points to
assigned based on at least one of size and radar cross section 40 form a three-dimensional radar image of the target.
of the target. 21. A multistatic radar tracking system comprising:
12. The method of claim 9, wherein the target type is at least one radar transmitter and a plurality of radar
assigned based on a change in radar cross section when the receivers arranged in node pairs, said at least one
three-dimensional image is viewed from different directions. transmitter emitting radar signals composed of ultra
13. The method of claim 1, comprising estimating a range 45 short wide bandwidth radio frequency pulses;
for a likely bistatic angle for the probable target data from
forward-scattering and backscattering radar cross sections. a filter for identifying at the receiver level from a radar
14. The method of claim 1, comprising recording a target signal received at the receiver of a node pair a target
map with the first target and at least one additional target. ellipse for a potential target; and
15. A method for identifying a target by using a multistatic 50 a real-time radar signal processing engine that determines
radar system having at least one radar transmitter and a from at least two target ellipses a first target location,
plurality of radar receivers forming node pairs, the method and determines an additional target location after
comprising: removing the target ellipses used to determine the first
receiving target data from a target in at least two receivers target location.
defining node pairs, the received target data defining a 55
target ellipse at each of the node pairs;