Introduction to Radar Systems

Tracking and Parameter Estimation

MIT Lincoln Laboratory

EST&TRACK-1
NBP 7/31/2008
 Disclaimer of Endorsement and Liability

• The video courseware and accompanying viewgraphs presented on this

server were prepared as an account of work sponsored by an agency of the
United States Government. Neither the United States Government nor any
agency thereof, nor any of their employees, nor the Massachusetts Institute
of Technology and its Lincoln Laboratory, nor any of their contractors,
subcontractors, or their employees, makes any warranty, express or
implied, or assumes any legal liability or responsibility for the accuracy,
completeness, or usefulness of any information, apparatus, products, or
process disclosed, or represents that its use would not infringe privately
owned rights. Reference herein to any specific commercial product,
process, or service by trade name, trademark, manufacturer, or otherwise
does not necessarily constitute or imply its endorsement, recommendation,
or favoring by the United States Government, any agency thereof, or any of
their contractors or subcontractors or the Massachusetts Institute of
Technology and its Lincoln Laboratory.

• The views and opinions expressed herein do not necessarily state or reflect
those of the United States Government or any agency thereof or any of their
contractors or subcontractors

EST&TRACK-2
MIT Lincoln Laboratory
NBP 7/31/2008
 Generic Radar Block Diagram

Waveform
Transmitter
Generator

Signal Processor
Target
Pulse Doppler
Antenna Receiver A/D
Compression Processing

Main Computer
Console /
Parameter Display
Detection Estimation
& Tracking
Recording

This lecture

EST&TRACK-3
MIT Lincoln Laboratory
NBP 7/31/2008
 Tracking Radars

BMEWS ASR
MOTR Courtesy Lincoln Laboratory
Courtesy of Lockheed Martin. Courtesy of Raytheon.
Used with permission. Used with permission.

Courtesy Lincoln Laboratory

Courtesy of US Navy.
AEGIS TRADEX
EST&TRACK-4
MIT Lincoln Laboratory
NBP 7/31/2008
 Parameter Estimation and Tracking
Functions

• After a target is initially detected, the radar must:

– Continue to detect the target
– Estimate target parameters from radar observations
Position, size, motion, etc.
– Associate detections with specific targets
Are all these nearby detections from the same target?
Use range, angle, Doppler measurements
– Predict where the target will be in the future
– Use multiple observations to develop a more accurate
filtered estimate of the target track

EST&TRACK-5
MIT Lincoln Laboratory
NBP 7/31/2008
 Outline

• Introduction

• Estimation
– Range Estimation

– Angle Estimation
Monopulse

– Estimation Performance

– Velocity (Doppler) Estimation

• Tracking

• Summary

EST&TRACK-6
MIT Lincoln Laboratory
NBP 7/31/2008
 Radar Parameter Estimation

• Location
– Azimuth Angle
– Elevation Angle
– Range
• Size
– Amplitude (RCS)
– Radial Extent (Length)
– Cross Range Extent (Width)
• Motion
– Radial Velocity
– Radial Acceleration
– Rotation, Precession
Radar – Ballistic Coefficient
EST&TRACK-7
EST&TRACK-5
MIT Lincoln Laboratory
NBP
NBP 7/31/2008
7/31/2008
 Parameter Estimation

• Primary metric parameters are range, angle, and Doppler velocity

Parameter Resolution Key Characteristics

Range 1 / BW Bandwidth

Angle λ/D Antenna size

Velocity
λ / Δt Coherent Integration Time
(Doppler)

– Accuracy improves as signal to noise ratio (SNR) increases

Resolution
σ ∝
SNR
• Basic approach: Overlapped measurements
– Range splitting
– Monopulse techniques
– Doppler bin splitting

EST&TRACK-8
MIT Lincoln Laboratory
NBP 7/31/2008
 Range Estimation

Output of Pulse Compression

True target location

• Range estimation uses

x multiple time samples for
peak fitting to achieve
greater accuracy
x
Voltage

• Range estimation accuracy

x x = time samples
improves with increasing
bandwidth
x
x 1 . 1
• Range accuracy ∝
x BW SNR
x x
x
x
Time
1 / BW

EST&TRACK-9
MIT Lincoln Laboratory
NBP 7/31/2008
 Increased Antenna Size Improves
Beamwidth
• Ability to resolve target directly impacts ability to
estimate target location

Parabolic Reflector Antenna Beamwidth

Antenna vs. Diameter
20
18 λ = 100 cm (300 MHz)

Half-Power Beamwidth (deg)

16 λ = 30 cm (1 GHz)
Feed λ = 10 cm (3 GHz)
14
D
12
10
8
Dish Decreasing
6 Wavelength

λ 180 4
~ D . π
Beamwidth (deg) ~
2

where D = aperture diameter 0

1 2 3 4 5 6 7 8 9 10
λ = wavelength Aperture Diameter D (m)

EST&TRACK-10
MIT Lincoln Laboratory
NBP 7/31/2008
 Angle Estimation

Antenna
Mainbeam
Response
• Detection provides coarse location in
1 2 3 4 5 angle
– Isolated within beamwidth of antenna
Antenna Response

• Typically greater accuracy is required

– 1º beam at 100 km extends across 1,745
meters!

• Angle Estimation uses measurements at

Angle different beam positions for greater
accuracy

1 2 3 4 5

EST&TRACK-11
MIT Lincoln Laboratory
NBP 7/31/2008
 Sequential Lobing Radar

Target Beam Responses

Beam 1 Axis
Beam 1 Beam 2

Beam 2

Power

Angle
Radar
• Time sequence of beams directed around track location
(two shown above)
• Reuses single receiver hardware for multiple beams
• Control loop redirects track location to equalize the beam
response

EST&TRACK-12
MIT Lincoln Laboratory
NBP 7/31/2008
 Conical Scan Tracking

Conical scan
Target Axis
follows axis of
Rotation Axis rotation

Beam
Rotation

Radar • Phase of modulation gives the angle

error

•
Amplitude

Amplitude of modulation gives the

Received

beam displacement

Time

EST&TRACK-13
MIT Lincoln Laboratory
NBP 7/31/2008
 Monopulse Angle Estimation

• Monopulse angle estimation

compares two or more simultaneous Monopulse Feed with Center Feed
receive beams
• The sum and difference of the two
squinted beams are used to generate
the error signal
– Each channel requires a separate
receiver
• Monopulse improves performance
over conical scan and sequential Courtesy Lincoln Laboratory
lobing whose performance degrade
with time varying radar returns
• Monopulse measurements can be
made via two methods
Multiple Simultaneous Receive Beams
– Amplitude-comparison (more
commonly used)
– Phase-comparison

EST&TRACK-14
MIT Lincoln Laboratory
NBP 7/31/2008
 Amplitude Comparison Monopulse

• Method:
– Pairs of offset receive beams used to determine the location
of the target relative to the antenna boresight (error signal)
– Error signal used to re-steer the antenna boresight on to the
target
• Typically, two offset receive beams are generated by using
two feeds slightly displaced from the focus of a parabolic
reflector
• The sum and difference of the two squinted beams are
used to generate the error signal
– Each channel requires a separate receiver

Antenna Antenna
Dish Feed
For A B
Azimuth
Monopulse
Radar

EST&TRACK-15
MIT Lincoln Laboratory
NBP 7/31/2008
 Amplitude Comparison Monopulse

Beam A Beam B
Response [dB]

A B

Hybrid
Angle A+B Junction A-B

• Receive two beams directed at

slightly different angles
– Typical offset 0.3 x beamwidth
• Generate Sum and Difference
Signals
– Sum = Σ = A + B
Sum Channel Difference Channel
– Difference = Δ = A - B Response ( Σ ) Response ( Δ )

EST&TRACK-16
MIT Lincoln Laboratory
NBP 7/31/2008
 Hybrid Junctions Used in Monopulse Radar

Σ B

Magic - T B Δ

A Δ
Σ A

Hybrid
Ring Junction
or “Rat-Race”

B1
A BΣ 3 dB
Directional
Δ
B2 ΣΔ Coupler

EST&TRACK-17
MIT Lincoln Laboratory
NBP 7/31/2008
 Example of Hybrid Junction

Σ B

Hybrid
Ring Junction
A Δ
or “Rat-Race”

• A signal input at port A reaches output port Δ

by two separate paths, which have the same path length ( 3λ/4)
– The two paths reinforce at port Δ
• An input signal at port B reaches output port Δ through paths differing by
one wavelength ( 5λ/4 and λ/4)
– The two paths reinforce at port Δ
• Paths from A to Δ and B to Δ differ by 1/2 wavelength
– Signal at port A - signal at port B will appear at port Δ
• If signals of the same phase are entered at A and B, the outputs Σ and Δ
are the sum and difference.

EST&TRACK-18
MIT Lincoln Laboratory
NBP 7/31/2008
 Example of Hybrid Junction
Σ B

Hybrid
Ring Junction A Δ
or “Rat-Race”

• A signal input at port A reaches output port Δ

by two separate paths, which have the same path length ( 3λ/4)
– The two paths reinforce at port Δ
• An input signal at port B reaches output port Δ through paths differing by
one wavelength ( 5λ/4 and λ/4)
– The two paths reinforce at port Δ
• Paths from A to Δ and B to Δ differ by 1/2 wavelength
– Signal at port A - signal at port B will appear at port Δ
• If signals of the same phase are entered at A and B, the outputs Σ and Δ
are the sum and difference.

EST&TRACK-19
MIT Lincoln Laboratory
NBP 7/31/2008
 Example of Hybrid Junction

Σ B Σ B

Hybrid
Ring Junction A Δ A Δ
or “Rat-Race”

• A signal input at port A reaches output port Δ

by two separate paths, which have the same path length ( 3λ/4)
– The two paths reinforce at port Δ
• An input signal at port B reaches output port Δ through paths differing by
one wavelength ( 5λ/4 and λ/4)
– The two paths reinforce at port Δ
• Paths from A to Δ and B to Δ differ by 1/2 wavelength
– Signal at port A - signal at port B will appear at port Δ
• If signals of the same phase are entered at A and B, the outputs Σ and Δ
are the sum and difference.
EST&TRACK-20
MIT Lincoln Laboratory
NBP 7/31/2008
 Monopulse Equations

1 0.4

Sum

(| Δ | / | Σ |) cos φ
0.5 0.2

k
Amplitude

=
e
0 0

op
sl
-0.5 -0.2
Difference Unamb.
region
-1 -0.4
-5 0 5 -1 -0.5 0 0.5 1
Angle (deg) Degrees

• Σ = Sum channel
• Δ = Difference channel
• φ = phase offset between Sum and Difference
| Δ | cos φ
• Error Signal e =
|Σ|

The Error Signal is a measure of how far the target is off-boresight

EST&TRACK-21
MIT Lincoln Laboratory
NBP 7/31/2008
 Two Dimensional Monopulse

Antenna Antenna
Dish Feed
B D
A C
Radar
• Σ = Sum channel signal
Δ = Difference channel signal
φ = phase difference between
Σ and Δ Note that the lower
| Δ | cos φ
• Error signal e =
|Σ|
feeds generate the
upper beams

Sum beam Elevation difference beam Azimuth difference beam

Σ Δ EL Δ AZ
B D B D B D
A C A C A C
A+B+C+D B+D – (A+C) B+A – (C+D)

EST&TRACK-22
MIT Lincoln Laboratory
NBP 7/31/2008
 Phase Comparison Monopulse

θ • Phase comparison monopulse also

known as “interferometer radar”
nts
vef
r o • Two antennas receive from the same
Wa
target direction
– Unlike amplitude comparison
monopulse that receives beams in
different directions
d sin(θ)
• Received target echo varies in phase
Antenna
– Δφ = 2π (d/λ) sinθ
d

EST&TRACK-23
MIT Lincoln Laboratory
NBP 7/31/2008
 Angle Estimation with Scanning Radar
(Multiple Pulse Angle Estimation)
Antenna Pattern Antenna Pattern
Target Target
Antenna
Pattern
(e.g. azimuth)

Power
Power

Detection
Threshold

Scan Angle Scan Angle

Scanned Output Power
Declared
Target

Power Detection
Threshold

Courtesy Lincoln Laboratory

Airport Surveillance Radar Scan Angle

EST&TRACK-24
MIT Lincoln Laboratory
NBP 7/31/2008
 Angle Estimation with Scanning Radar
(Multiple Pulse Angle Estimation)
Antenna Pattern Antenna Pattern
Target Target

Power
Power

Antenna Detection
Pattern Threshold

Scan Angle Scan Angle

Scanned Output Power
• For a “track-while scan” radar, Declared
the target angle is measured by: Target
– the highest target return, or
– Interpolated angle Power Detection
measurement using known Threshold
antenna pattern

Scan Angle

EST&TRACK-25
MIT Lincoln Laboratory
NBP 7/31/2008
 Angle Estimation with Array Antennas

• Phased array radars are well suited for monopulse

tracking
– Amplitude Comparison Monopulse
Radiating elements can be combined in 3 ways
Sum, azimuth difference, and elevation difference
patterns
– Phase Comparison Monopulse BMEWS
Use top and bottom half of array for elevation Courtesy of Raytheon.
Used with permission.
Use right and left half of array for azimuth
• Lens arrays (e.g. MOTR) would use amplitude
monopulse
– Four-port feed horn would be same as for dish
reflector

MOTR
Courtesy of Lockheed Martin.
Used with permission.

EST&TRACK-26
MIT Lincoln Laboratory
NBP 7/31/2008
 Monopulse Angle Estimation Accuracy

SNR = 13 dB
0.8
Angle Error/Beamwidth

0.6

Angle Accuracy ∝ Beamwidth

0.4 SNR

0.2

0 10:1 Beam splitting

-10 0 10 20 30
Signal-to-noise ratio [dB]

At
Attypical
typicaldetection
detectionthreshold
thresholdlevels
levels(~13
(~13dB)dB)the
thebeamwidth
beamwidthcan
canbebe
approximately
approximatelysplit
splitby
byaafactor
factorof
often;
ten;i.e.
i.e.10:1
10:1antenna
antennabeam
beamsplitting
splitting

EST&TRACK-27
MIT Lincoln Laboratory
NBP 7/31/2008
 Accuracy, Precision and Resolution
• Accuracy: Example
– The degree of conformity of measurement to Accuracy vs. Precision
the true value
• Precision:
– Repeatability of a measurement
– Bias Error : True value- Average measured
value
• Resolution: Low Accuracy High Accuracy Low Accuracy
– Offset (angle or range) required for two Low Precision High Precision High Precision
targets to be recognized as separate targets

Targets at 0° and 6° Targets at 0° and 3°

0 0
Amplitude (dB)

-5
Amplitude (dB)
-5
-10 -10
-15 -15
-20 -20
-25 -25
-30 -30
-10 0 10 20 -10 0 10 20
Angle (deg) Angle (deg)
EST&TRACK-28
MIT Lincoln Laboratory
NBP 7/31/2008
 Doppler Velocity Estimation

2vr Radial Velocity

Doppler Frequency fd =
λ Wavelength

0
• Use two closely spaced frequency
filters offset from the center
Filter Response [dB]

-10
frequency of the Doppler filter
-20 containing the detection
-30 • Velocity estimation procedure is
similar to angle estimation with
-40 angle and frequency interchanged
-50 • Doppler measurement accuracy
∝ λ . 1
0 Target velocity va
(ambiguous velocity)
Doppler Frequency Δt SNR
(Δt = coherent integration time)

EST&TRACK-29
MIT Lincoln Laboratory
NBP 7/31/2008
 Real-World Limitations

• Receiver noise
– Adds variance to estimates
• Radar calibration
– Poor calibration leads to poor estimation
• Amplitude fluctuations
– Small effect on monopulse and array solutions
• Angle noise (angle scintillations, or target glint)
– Complex target return biases angle estimate
• Multipath (low angle tracking)
– Reflection off earth’s surface combines with direct path return
– Can cause biases in angle estimates for all techniques

EST&TRACK-30
MIT Lincoln Laboratory
NBP 7/31/2008
 Real-World Limitations

• Receiver noise
– Adds variance to estimates
• Radar calibration
– Poor calibration leads to poor estimation
• Amplitude fluctuations
– Small effect on monopulse and array solutions
• Angle noise (angle scintillations, or target glint)
– Complex target return biases angle estimate
• Multipath (low angle tracking)
– Reflection off earth’s surface combines with direct path return
– Can cause biases in angle estimates for all techniques
Direct path

Radar
Multi-path
EST&TRACK-31
MIT Lincoln Laboratory
NBP 7/31/2008
 Outline

• Introduction

• Estimation

• Tracking

• Summary

EST&TRACK-32
MIT Lincoln Laboratory
NBP 7/31/2008
 Radar Tracking Example

Tracker Input Tracker Output

Cross-Range

Range Range
New Track
Observations
Existing Track
• Tracker receives new observations • New tracks are initiated
every scan • Existing tracks are updated
– Target observations
– False alarms
• Obsolete tracks are deleted

EST&TRACK-33
MIT Lincoln Laboratory
NBP 7/31/2008
 Automatic Detection and Tracking
Techniques

• Development of clutter rejection techniques and the digital

revolution have enabled the successful development of these
automatic detection and tracking techniques for Air Defense
and Air Traffic Control radar systems
• Detection and Tracking Functions
– Target Detection
Adaptive threshold (CFAR) applied to each range, angle,
Doppler cell
– Target Association
Adjacent ( range, angle, and Doppler) threshold crossings, are
associated
Range, angle(s), and Doppler of target are calculated from
associated detections

EST&TRACK-34
MIT Lincoln Laboratory
NBP 7/31/2008
 Tracking Tasks

New Observations
• Track association and Detection Reports:
update range, range rate,
– Attempt made to correlate azimuth, elevation,
new detection with an etc.
existing tracks
– Association is aided by Non-associated
seeing if the detections fall Observations
within a search window Association
• Track initiation
– Track initiated from several Existing
scans of detection Initiation Tracks Prediction
information
– Track initiation in dense
clutter environment can Update
stress computer resources New
Tracks

Termination

EST&TRACK-35
MIT Lincoln Laboratory
NBP 7/31/2008
 Tracking Tasks
Maneuvering Gate Non-maneuvering New Observations
Gate
Detection Reports:
Present Target range, range rate,
Predicted Location azimuth, elevation,
Location etc.
Next Scan
• Track association and update Non-associated
– The size of the gate is determined by Observations
Association
Estimated errors in the predicted
position
Estimated errors in the speed and Existing
direction of the track Initiation Tracks Prediction
– The gate should be :
Small in order to avoid having more
than one detection fall within the gate Update
New
Large to follow target turns or Tracks
maneuvers
– If target association is successful,
the track files are updated with the
Termination
new target detection data

EST&TRACK-36
MIT Lincoln Laboratory
NBP 7/31/2008
 Tracking Tasks

New Observations
• Track prediction (filtering)
Detection Reports:
– Past detections used to estimate the range, range rate,
target’s present position and velocity azimuth, elevation,
– Estimate used to predict the location etc.
of the target on the next scan
– Different methods of smoothing the
Non-associated
detection data Observations
ƒ α−β Filter Association
ƒ Kalman Filter
Existing
Initiation Prediction
• Track termination Tracks
– If data from target is missing on a
scan of radar, track may be
Update
“coasted” New
– If data from target missing for a Tracks
number of scans, the track is
terminated
Termination

EST&TRACK-37
MIT Lincoln Laboratory
NBP 7/31/2008
 Tracking with Phased Array Radar

• Tracking techniques are similar to automatic detection and

tracking just described
• Advantages of phased array
– Higher track update rate than radars with mechanically scanned
antennas
– Can simultaneously track multiple targets separated by many
beamwidths
• There is no closed loop feedback controlling the radar beam
– Computer controls the radar beam and track update rate

Courtesy of U. S. Navy. Courtesy of Raytheon. Used with permission.

EST&TRACK-38
MIT Lincoln Laboratory
NBP 7/31/2008
 Track Before Detect Techniques

• Probability of detection may be improved by non-coherently

integrating the radar echoes over multiple scans of the
radar
– Long integration times implies target may traverse many
resolution cells during the integration time
– Since target trajectory usually not known beforehand,
integration must be performed assuming all possible
trajectories
Computationally intensive problem
– A correct trajectory is one that provides a realistic speed and
direction for the type of target being observed
– The target must be tracked before it is detected
Also called: Retrospective detection, long term integration
– Higher single scan probability of false alarm can be tolerated
PFA= 10-3 rather than 10-5 or 10-6
– Requires :
Increased data processing capability
Longer observation time

EST&TRACK-39
MIT Lincoln Laboratory
NBP 7/31/2008
 Summary

• Parameter estimation techniques enable a radar to obtain

accurate radar measurements
– Range, angle, Doppler, etc.

• Monopulse angle estimation allows sub-beamwidth

accuracy for a single radar pulse
– Limitations due to multiple targets or interference

• Tracking algorithms find best fit between predicted target

track and current observations

EST&TRACK-40
MIT Lincoln Laboratory
NBP 7/31/2008
 References

• Skolnik, M., Introduction to Radar Systems, New York,

McGraw-Hill, 3rd Edition, 2001
• Toomay, J. C., Radar Principles for the Non-Specialist, New
York, Van Nostrand Reinhold, 1989

EST&TRACK-41
MIT Lincoln Laboratory
NBP 7/31/2008