Radar Systems Engineering Lecture 13 Clutter Rejection

Part 2 - Doppler Filtering

Dr. Robert M. ODonnell IEEE New Hampshire Section Guest Lecturer

IEEE New Hampshire Section

Radar Systems Course 1 Pulse Doppler 11/1/2009

IEEE AES Society

Block Diagram of Radar System

Transmitter Power Amplifier Waveform Generation

Propagation Medium Target Radar Cross Section T/R Switch Antenna

Signal Processor Computer Receiver

User Displays and Radar Control

A/D Converter

Pulse Compression

Clutter Rejection (Doppler Filtering)

General Purpose Computer Parameter Estimation

Tracking Data Recording

Photo Image Courtesy of US Air Force

Thresholding

Detection

Radar Systems Course 2 Pulse Doppler 11/1/2009

IEEE New Hampshire Section IEEE AES Society

Outline

Introduction Problem perspective
Burst Waveforms and their properties The impact of Moores Law on radar Signal Processing
Past, present, and the future

Pulse Doppler Processing Techniques

Description of pulse Doppler processing Low PRF Example Moving Target Detector (MTD) Range and Doppler Ambiguities Ambiguity Resolution - Chinese remainder theorem The Ambiguity Function Preview of Airborne Pulse Doppler issues

Radar Systems Course 3 Pulse Doppler 11/1/2009

Summary
IEEE New Hampshire Section IEEE AES Society

The Problem- Not Just Ground Clutter

Ground Clutter

Sea Clutter
Less intense than ground
By 20 to 30 dB Often more diffuse

Can be intense and discrete Can be 50 to 60 dB > than target Doppler velocity zero for ground based radars
Doppler spread small

Doppler velocity varies for based radars (ship speed & wind speed)
Doppler spread moderate

Rain Clutter
Diffuse and windblown Can be 30 + dB > than target Doppler zero for ground based radars
Doppler spread small Doppler Spectrum of Rain
Received Power (dB)
80 60 40 20 0 60 Kt 0 + 60 Kt

Bird Clutter
100s to 10,000s to point targets Doppler velocity - 0 to 60 knots
Doppler of single bird has little change Flocks of birds can fill 0 to 60 knots of Doppler space Big issue for very small targets

A one filter with a notch at zero Doppler will not adequately reject rain

Doppler Velocity
Radar Systems Course 4 Pulse Doppler 11/1/2009

Courtesy of FAA

IEEE New Hampshire Section IEEE AES Society

Issues with MTI Cancellers

Typically they process a few (2-5) pulses at a time, so it is near to impossible to shape them as well as you could if filter had an input of 8-10 pulses
2 pulse MTI canceller is very broad in Doppler space

A set of pass band Doppler filters, using 8-10 pulses) can be constructed having:
A notch at zero Doppler to reject ground clutter A set of passband filters that can detect targets where no rain is present

Before the mid 1970s, the technology, to cost effectively implement pulse Doppler solutions to the simultaneous ground and rain clutter was not available

Radar Systems Course 5 Pulse Doppler 11/1/2009

IEEE New Hampshire Section IEEE AES Society

Outline

Introduction Problem perspective
Burst Waveforms and their properties The impact of Moores Law on radar Signal Processing
Past, present, and the future

Pulse Doppler Processing Techniques

Description of pulse Doppler processing Low PRF Example Moving Target Detector (MTD) Range and Doppler Ambiguities Ambiguity Resolution - Chinese remainder theorem The Ambiguity Function Preview of Airborne Pulse Doppler issues

Radar Systems Course 6 Pulse Doppler 11/1/2009

Summary
IEEE New Hampshire Section IEEE AES Society

Utility of Burst Waveforms for Clutter Rejection

Pulse 1 Sample 13 e.g. 4.6 km

Pulse 2 Sample 13 e.g. 4.6 km

Pulse 3 Sample 13 e.g. 4.6 km

Time Range Samples at 13th range gate for Mth pulse

A burst waveform offers a method of collecting M sequential samples an each range - CPI cell. These samples can be linearly processed through a set of pass band filters that will
Detect targets within a range of Doppler velocities and simultaneously reject clutter that is in their low sidelobes If the pass band filters are narrow enough in frequency, a measurement of the Doppler velocity of the target that passes through them can be made
IEEE New Hampshire Section IEEE AES Society

Radar Systems Course 7 Pulse Doppler 11/1/2009

Outline

Introduction Problem perspective
Burst Waveforms and their properties The impact of Moores Law on Radar Signal Processing
Past, and present, and the future

Pulse Doppler Processing Techniques

Description of pulse Doppler processing Low PRF Example Moving Target Detector (MTD) Range and Doppler Ambiguities Ambiguity Resolution - Chinese remainder theorem The Ambiguity Function Preview of Airborne Pulse Doppler issues

Radar Systems Course 8 Pulse Doppler 11/1/2009

Summary
IEEE New Hampshire Section IEEE AES Society

Impact of Moores Law of Radar Processing

Tremendous advances in A/D Converter technology In the 1970s, a 10 bit 5 MHz A//D was near the commercial state of the art
30 lbs and 3 of rack space Now it is not only on a chip, but many more bits and much higher sample rates are available

For a 60 nmi aircraft surveillance radar, with a mechanically scanning antenna, the new computational processing advances allowed the number of range-azimuth-Doppler cells being individually thresholded from a several thousand to several million per radar scan
These advances allowed aircraft to be reliably detected in rain Much better detection of aircraft in ground clutter Low false alarm rates that allowed the radar and beacon sensor systems to be seamlessly integrated

In the future, expect that advances in processing technology will allow, implementation of new techniques, which today are seemingly impossible to implement
IEEE New Hampshire Section IEEE AES Society

Radar Systems Course 9 Pulse Doppler 11/1/2009

Outline

Introduction Problem perspective
Burst Waveforms and their properties The impact of Moores Law on radar Signal Processing
Past, present, and the future

Pulse Doppler Processing Techniques

Description of Pulse Doppler processing Low PRF Example Moving Target Detector (MTD) Range and Doppler Ambiguities Ambiguity Resolution - Chinese remainder theorem The Ambiguity Function Preview of Airborne Pulse Doppler issues

Radar Systems Course 10 Pulse Doppler 11/1/2009

Summary
IEEE New Hampshire Section IEEE AES Society

Waveforms for Pulse Doppler Processing Revisited

TCPI = NTPRI
Radar Signal

TPRI T

Time

For Airport Surveillance Radar

T B = = 1/T Pulse Length Bandwidth Pulse Repetition Interval (PRI) Pulse Repetition Frequency (PRF) Duty Cycle (%) 1 sec 1 MHz 1 msec 1 KHz .1 %

TPRI = fP = 1/TPRI = T /TPRI TCPI = NTPRI N =

Coherent Processing Interval (CPI) 10 pulses Number of pulses in the CPI N = 2, 3, or 4 for MTI N usually much greater (8 to ~1000) for Pulse Doppler
IEEE New Hampshire Section IEEE AES Society

Radar Systems Course 11 Pulse Doppler 11/1/2009

Data Collection for Doppler Processing

Pulse 1 Sample 13 e.g. 4.6 km

Pulse 2 Sample 13 e.g. 4.6 km

Pulse 3 Sample 13 e.g. 4.6 km

Time Range Samples at 13th range gate for Mth pulse M

A/D Converter

(the complex envelope of received waveform)

Pulse Number (Slow time) 1

I&Q Samples (Real and Imaginary)

Complex I / Q samples

What is the optimum way to process the M voltage samples in a range gate to detect the target and reject clutter?
Radar Systems Course 12 Pulse Doppler 11/1/2009

Sample No. Range

Viewgraph Courtesy of MIT Lincoln Laboratory Used with permission

IEEE New Hampshire Section IEEE AES Society

Pulse Doppler Processing - Cartoon

Frequency Response of the M Filters

Doppler Filter Bank

r S = {s1 , s 2 , s 3 , ... , s M }
M radar echo samples for each range CPI cell

Filter 1 (f1, v1) Filter 2 (f2, v2) Filter 3 (f3,v3) M Doppler velocity bins out Filter M (fM,vM)

Doppler Frequency Coherent integration of all pulses of a CPI Doppler Velocity Clutter rejection Resolving targets into different velocity segments and allowing for fine-grain target radial velocity estimation

Viewgraph Courtesy of MIT Lincoln Laboratory Used with permission

Radar Systems Course 13 Pulse Doppler 11/1/2009

IEEE New Hampshire Section IEEE AES Society

Optimization of Pulse Doppler Processing

MTI Improvement Factor, I , already introduced in previous lecture is

I (fd) =

(Signal / Clutter)out (Signal / Clutter)in

The next question In the presence of colored noise (ground clutter, rain & noise), what are the weights, Wi (f D ) , by which the M input signal (+ clutter) samples, S i , should by multiplied by so that the S/(C+N) will be maximized?
Note that the optimum set on weights depends on f D Also on the number of pulses processed, M

In the late 1960s, the solution was developed by 2 independent sets of researchers (See Reference14 and 15)

Radar Systems Course 14 Pulse Doppler 11/1/2009

IEEE New Hampshire Section IEEE AES Society

Optimum MTI Improvement Factor

Problem
What is the optimum way (maximize S /( N + C + I ) to linearly process M complex radar echoes,Vi , in the presence of noise, clutter returns (ground, rain, sea, etc.) and interference?

Answer:

R=

WV
i =1 i
M j= 1

I OPT = Sk WkOPT
1 ij

where W

OPT i

= k M Sj

Vi
M ij Si
k

= Sampled voltage (sum of target echo, clutter, noise, etc.) = Covariance matrix of clutter, noise, etc = Signal vector = arbitrary constant = Number of pulses processed
See De Long et al, Reference 14 for detailed derivation

M
Radar Systems Course 15 Pulse Doppler 11/1/2009

IEEE New Hampshire Section IEEE AES Society

Pulse Doppler Processing

The optimum weights are given by:

OPT i

= k M ij1S j

j= 1

The optimum filter weights are a function of Doppler frequency

In lecture 18, these issues will be studied in more detail.

Also, see Reference 9
Its a great, instructive readable reference for this material

Because of the variable nature of ground clutter and rain, a simple high pass filter (MTI canceller) using a few (2-5) pulses will not come close to simultaneously rejecting both ground and rain clutter
At least 8 to 10 pulses are required for good rain rejection Much of the rain clutter will pass through a high pass filter

Typically, a set (bank) of Doppler filters are used, in parallel, to given good target detection over the range of Doppler frequencies
0 to the PRF (Blind Speed) The number of filters usually is equal to the number of pulses processed
IEEE New Hampshire Section IEEE AES Society

Radar Systems Course 16 Pulse Doppler 11/1/2009

MTI Improvement Factor Comparison

60 10 Pulse Optimum Processor MTI Improvement Factor
Typical Airport Surveillance Radar

40 3 Pulse Canceller
Courtesy of Frank Sanders

20

Wavelength S-Band 10.7 cm Antenna width 17.5 ft Rotation rate 13 rpm PRF 1000Hz C/N =40 dB

0 0
Radar Systems Course 17 Pulse Doppler 11/1/2009

500 Doppler Frequency (Hz)

1000
IEEE New Hampshire Section IEEE AES Society

MTI Improvement for One Optimum Filter

Filter optimized to reject ground clutter and noise 60 Envelope of all optimum filters MTI Improvement Factor Optimum filter tuned to 300 Hz
Typical Airport Surveillance Radar

40

20
Courtesy of Frank Sanders

0 0

200

400 600 800 Doppler Frequency (Hz) 55 Radial Velocity (knots)

1000 110

Wavelength S-Band 10.7 cm Antenna width 17.5 ft Rotation rate 13 rpm PRF 1000Hz C/N =40 dB No. of pulses 10

Radar Systems Course 18 Pulse Doppler 11/1/2009

IEEE New Hampshire Section IEEE AES Society

Implementations of a Set of Doppler Filters

Doppler Filter Bank (linear scale) 8 Filters - DFT- 13 dB Sidelobes
Magnitude (linear scale) Magnitude (dB)

Doppler Filter Bank 8 Filters - Shaped - Low Doppler Sidelobes

10 0 -10 -20 -30 -40 -50 -60

0.25 0.5 0.75 Doppler Frequency (f/PRF)

1.0

-70

0.25 0.5 0.75 Doppler Frequency (f/PRF)

1.0

The simplest way to implement a bank of filters is with a Discrete Fourier Transform (DFT)
Note the 13 dB sidelobes will give poor suppression of rain clutter Weighting the input signal or use of other techniques, to be discussed in the next lecture, along with integrating an adequate number of pulses will give excellent target detection in the presence in even heavy rain Viewgraph Courtesy of MIT Lincoln Laboratory
Used with permission

Radar Systems Course 19 Pulse Doppler 11/1/2009

IEEE New Hampshire Section IEEE AES Society

Outline

Introduction Problem perspective
Burst Waveforms and their properties The impact of Moores Law on radar Signal Processing
Past, present, and the future

Pulse Doppler Processing Techniques

Description of pulse Doppler processing Low PRF Example Moving Target Detector (MTD) Range and Doppler Ambiguities Ambiguity Resolution - Chinese remainder theorem The Ambiguity Function Preview of Airborne Pulse Doppler issues

Radar Systems Course 20 Pulse Doppler 11/1/2009

Summary
IEEE New Hampshire Section IEEE AES Society

Aircraft Surveillance Radar (ASR) Problem

Courtesy of MIT Lincoln Laboratory Used with permission

ASR-9

Range 60 nmi.

Range Resolution 1/16 nmi Rotation Rate 12. rpm Transmits Pulses at ~ 1250 Hz Az Beamwidth ~1.2

Range - Azimuth - Doppler Cells

~1000 Range cells ~500 Azimuth cells ~8-10 Doppler cells 5,000,000 Range-Az-Doppler Cells to be threshold every 4.7 sec.

As Antenna Rotates ~22 pulses / Beamwidth

IEEE New Hampshire Section IEEE AES Society

Radar Systems Course 21 Pulse Doppler 11/1/2009

Moving Target Detector (MTD) Processor

Issue
Ground Clutter

Solution
1. Eight Pulse Doppler Linear Coherent Filters (10 pulses) 2. Coherent Transmitter 3. Constant PRF within coherent processing interval 4. Doppler Filter Bank 5. Adaptive Thresholding for each (Range Azimuth Doppler) Cell - 3.9 million cell 6. Fine Grained Clutter Map 7. Multiple PRFs
IEEE New Hampshire Section IEEE AES Society

Second Time Around Clutter

Rain

Tangential Targets, Blind Speeds

Radar Systems Course 22 Pulse Doppler 11/1/2009

Moving Target Detector (MTD)

I and Q Digitized Radar Echoes From Each Range Cell

8 or Greater Pulse Doppler Filter Bank Zero Velocity Filter

Adaptive Thresholding

Post Processing Thresholding

Output Detections

Clutter Map Filter

Pulse Doppler filtering on groups of 8 or more pulses with a fine grained clutter map. Aircraft are detected in ground clutter and / or rain with the Doppler filter bank & use of 2 PRFs. Birds and ground traffic are rejected in post processing, using Doppler velocity and a 2nd fine grained clutter map
IEEE New Hampshire Section IEEE AES Society

Viewgraph Courtesy of MIT Lincoln Laboratory Used with permission

Radar Systems Course 23 Pulse Doppler 11/1/2009

ASR-9 8-Pulse Filter Bank

Note: Doppler sidelobes 40 dB down from peak response

Courtesy of MIT Lincoln Laboratory

ASR-9 Filter Bank 10 Magnitude (dB) 0 Magnitude (dB) -10 -20 -30 -40 -50 -60 -70 0 20 40 60 80 Radial Velocity (kts) 100 10 0 -10 -20 -30 -40 -50 -60 -70 0

ASR-9 One Doppler Filter

Rain Echo

Aircraft

20

40 60 80 Radial Velocity (kts)

100

Viewgraph Courtesy of MIT Lincoln Laboratory Used with permission

Radar Systems Course 24 Pulse Doppler 11/1/2009

IEEE New Hampshire Section IEEE AES Society

Detection in Rain Using Two PRFs

Filter Number 4 PRF 1 4 PRF 2 5 6 7 0 1 2 3 4 5 6 7 0 1 2 3 4

Aliased Values of Aircraft Doppler Velocity At Different PRFs

True Aircraft Velocity 0

Radial Velocity
IEEE New Hampshire Section IEEE AES Society

Radar Systems Course 25 Pulse Doppler 11/1/2009

MTD Performance in Rain

Unprocessed Radar Returns Time History of MTD Radar Tracker Output August 1975, FAA Test Center

Doppler Spectrum of Rain

Received Power (dB)
80 60 40 20 0 60 Kt 0 + 60 Kt

Doppler Velocity

Photographs Courtesy of FAA

Radar Systems Course 26 Pulse Doppler 11/1/2009

IEEE New Hampshire Section IEEE AES Society

Moving Target Detector - I (1975)

A/D Converters
10 Bit 2.6 MHz Top one Q Bottom one I

Input Memory
Corner turning memory

MTD Processor
~1000 TTL Chips

Clutter Map
Using Drum memory technology

Courtesy of FAA

Analog IF Chassis
IEEE New Hampshire Section IEEE AES Society

Radar Systems Course 27 Pulse Doppler 11/1/2009

Non-Coherent Integration and the Effect of Correlated Clutter

Doppler Spectrum of Rain
Received Power (dB)
80 60 40 20 0 60 Kt
Courtesy of FAA

+ 60 Kt

Doppler Velocity

Rain clutter residue that leaks through the MTI canceller is correlated from pulse to pulse Non coherent integration of correlated clutter residue is less efficient than with uncorrelated noise

Non Coherent Integration Probability Distribution Single Pulse Uncorrelated Correlated

Sum of 10 Pulse

Radar Systems Course 28 Pulse Doppler 11/1/2009

IEEE New Hampshire Section IEEE AES Society

Equivalent number of independent pulses

Correlation Function (T) 0.99 100 50 Dependent Sampling Region

Independent Pulses for Partially Correlated Waveforms

0.97 0.9 0.7 0.5 0.15 0.02

N Number of pulses

V
30 0
Standard Deviation of Clutter Velocity

20 10

50

0 10

Wavelength

N
5

20

30

f PRF = 1 / T
10
Pulse Repetition Rate

2 1 0.001 0.002

=3 N

Blue dot is for an S band 2D radar in rain clutter

Independent Sampling Region 0.2 0.5 1.0
Adapted from Nathanson, Reference 5

0.005

0.01

V / f PRF

0.02

0.05

0.1

Non-coherent integration of partially correlated pulses can often be very inefficient

IEEE New Hampshire Section IEEE AES Society

Radar Systems Course 29 Pulse Doppler 11/1/2009

MTD Implementations

The first 2 versions of the MTD were designed, built for the FAA by MIT Lincoln Laboratory from the early to the late 1970s and are documented extensively* After operational testing of MTD II, These concepts were included into the specification of the ASR-9 and incorporated in that radar** along with additional improvements These concepts are presently implemented in almost all ground based low PRF radars and have influenced the extensive evolution of pulse Doppler processing onto sea based and airborne platforms, as further improvements in digital processing technology and algorithmic techniques have advanced

*See References 1, 8, 9 , 10, 11, 12 for extensive discussion **See Taylor References 15
Radar Systems Course 30 Pulse Doppler 11/1/2009

IEEE New Hampshire Section IEEE AES Society

MTD Summary

The MTD proved that, for the first time, digital signal processing hardware and algorithmic technology could be implemented in a manner that would give excellent aircraft detection while rejecting all forms of clutter (ground, rain, etc), under almost all conditions, so that radar and beacon reports could be reliably correlated and displayed to the air traffic controller. Solving this particular civilian problem has been, over the ensuing years, a catalyst, for the appropriate application of this general approach to many other civilian and military radar problems:
Understanding that Moores law will allow cost effective use, in the near future, of processing techniques, seemingly not cost effective today
Some experts said You can never make wire wrapped 1000 IC work reliably ( Incidentally, they were wrong!) Now that processing can be done with a few programmable Power PC cards

Integration of many pulses to use low Doppler sidelobes to reject moving clutter (rain, chaff , sea clutter, etc.) Use of high resolution clutter maps, to detect tangential targets Solving the signal processing to radar target display problem in an integrated manner
IEEE New Hampshire Section IEEE AES Society

Radar Systems Course 31 Pulse Doppler 11/1/2009

MTD Clutter Map Techniques

Clutter maps are a memory which stores for each range-CPI cell in the radars coverage the value of the noise and clutter echo in that cell
Clutter maps are usually implemented using a recursive filter For each range CPI cell, the clutter map is updated using the following algorithm

A(n + 1) =

1 (A(n)) + 1 1 (A(n 1)) N N

n = scan number

N = 8 for the MTD

They are used to detect targets whose radial velocity is at of near zero and whose backscatter echo is greater than the clutter and / or noise amplitude stored in the clutter map
The clutter map channel offers a method of detecting targets that are not detected by the subset of the Doppler filters, that are adjacent to zero Doppler and whose shape is designed to strongly reject ground echoes near zero Doppler

Radar Systems Course 32 Pulse Doppler 11/1/2009

IEEE New Hampshire Section IEEE AES Society

Clutter Map Thresholding

Clutter map detection techniques use temporal thresholding techniques

Spatial CFAR techniques would detect the edges of moving rain clouds

Target detection is declared if the size of the average of the coherently integrated return is M times the previous scans value, which is stored on the clutter map This process is performed for each Range CPI cell every scan of the radar
~350,000 cell for an ASR radar

Additional Points
This technique makes possible detection of tangential aircraft flying tangentially near large discrete pieces of ground clutter
Called Inter-clutter visibility in the literature

Aircraft moving tangentially to the radar are give large specular echoes, which enhances this detection mode

Radar Systems Course 33 Pulse Doppler 11/1/2009

IEEE New Hampshire Section IEEE AES Society

Post Signal Processing Clutter Map Techniques

Even with these, relatively sophisticated signal processing and thresholding techniques, performed on single range CPI basis, sometimes excessive false detections do occur These can be caused by
Heavy bird migration Ground clutter whose echoes exceed the A/D dynamic range Automobile traffic And other sources

More sophisticated Area CFAR very similar to clutter maps have been developed to effectively deal with these problems
This set of thresholding techniques are employed before the tracking function Good places to learn more detail about these post processing techniques are detailed ;
References 11 and 12; Reference 6, pp 284-285
IEEE New Hampshire Section IEEE AES Society

Radar Systems Course 34 Pulse Doppler 11/1/2009

Outline

Introduction Problem perspective
Burst Waveforms and their properties The impact of Moores Law on radar Signal Processing
Past, present, and the future

Pulse Doppler Processing Techniques

Description of pulse Doppler processing Low PRF Example Moving Target Detector (MTD) Range and Doppler Ambiguities Ambiguity Resolution - Chinese remainder theorem The Ambiguity Function Preview of Airborne Pulse Doppler issues

Radar Systems Course 35 Pulse Doppler 11/1/2009

Summary
IEEE New Hampshire Section IEEE AES Society

Unambiguous Doppler Velocity and Range

Unambiguous Range (nmi)
400 3000 100
Hz Hz G

40

10

First Blind Speed (knots)

1000

300

100

d 5 an 43 B Hz F G nd 3 VH Hz Ba 1. G d 2 HF Hz U an G 3. B 4 d L 9. an Hz d B G n S 35 Ba d X an B a K

0 22

f PRF VB = 2
and

c RU = 2 f PRF
Yields

30

10 0.1

10

100

c VB = 4 RU

Pulse Repetition Rate (KHz)

IEEE New Hampshire Section IEEE AES Society

Radar Systems Course 36 Pulse Doppler 11/1/2009

Sensitivity Time Control

Aircraft at 200 nmi, RCS = 1 m2 Bird at 89 nmi, RCS = 0.0015 m2

These two targets have the same detectability, because in the radar equation: S
N R4

This false target issue can be mitigated by attenuating to the received signal by a factor which varies as 1/R4
Can also be accomplished by injecting noise into the receive channel , which falls off as 1/R4

Radars that utilize range ambiguous waveforms, cannot use STC, because long range targets which alias down in range, would be adversely attenuated by the STC
For these waveforms, other techniques must be used to mitigate the false target problem due to birds

Radar Systems Course 37 Pulse Doppler 11/1/2009

IEEE New Hampshire Section IEEE AES Society

Classes of MTI and Pulse Doppler Radars

Low PRF Range Measurement Velocity Measurement Low PRF Wind blown clutter may be a problem Can use STC
Unambiguous

Medium PRF
Ambiguous

High PRF
Very Ambiguous

Very Ambiguous

Ambiguous

Unambiguous

Medium PRF Wind blown clutter may be a problem Range eclipsing losses Far out targets compete with near in clutter Cant use STC Ambiguities difficult to remove

High PRF Range eclipsing losses Distant targets compete with near in clutter Cant use STC

Viewgraph Courtesy of MIT Lincoln Laboratory Used with permission

Radar Systems Course 38 Pulse Doppler 11/1/2009

IEEE New Hampshire Section IEEE AES Society

Outline

Introduction Problem perspective
Burst Waveforms and their properties The impact of Moores Law on radar Signal Processing
Past, present, and the future

Pulse Doppler Processing Techniques

Description of pulse Doppler processing Low PRF Example Moving Target Detector (MTD) Range and Doppler Ambiguities Ambiguity Resolution - Chinese remainder theorem The Ambiguity Function Preview of Airborne Pulse Doppler issues

Radar Systems Course 39 Pulse Doppler 11/1/2009

Summary
IEEE New Hampshire Section IEEE AES Society

Velocity Ambiguity Resolution

Unfold detections out to some maximum velocity

CPI #1 PRF = f1
f1 2f1 3f1 4f1 5f1

Blind Zones

CPI #2 PRF = f2
Individual CPI unambiguous velocity regions f2 2f2 3f2 4f2

Doppler (Velocity)

Doppler (Velocity)

Split dwell into multiple CPIs at different PRFs

Scan to scan, even pulse-to-pulse changes also possible

Moves blind velocities to ensure detection of all non-zero velocity targets True target velocity is where best correlation across CPIs occurs Choose PRFs so that least common multiple occurs above desired Viewgraph Courtesy of MIT Lincoln Laboratory Used with permission maximum unambiguous velocity
IEEE New Hampshire Section IEEE AES Society

Radar Systems Course 40 Pulse Doppler 11/1/2009

Chinese Remainder Theorem

Rc = ( C1A1 + C2A2 + C3A3 ) modulo (m1 m2 m3 )
(assumes 3 PRFs)

Rc = True range/Doppler cell number

Cell number is range expressed in pulse widths or Doppler velocity expressed in Doppler filter widths

Ai = Ambiguous range or Doppler cell number for ith PRF PRFi = 1 / t mi t = pulsewidth

m1 m2 m3 are relatively prime numbers C1 C2 and C3 are related to m1 m2 and m3 by C1 = b1 x m2 m3 = 1 modulo m1 C2 = b2 x m3 m1 = 1 modulo m2 C3 = b3 x m1 m2 = 1 modulo m3 where b1 = smallest positive integer which, when multiplied by m2 m3 and divided by m1 gives unity as the remainder
Radar Systems Course 41 Pulse Doppler 11/1/2009

IEEE New Hampshire Section IEEE AES Society

Example - Chinese Remainder Theorem

Doppler Filters at PRF1
1 2 3 4 5 6 7 8 1 2 3 4 5 6 7 8 1 2 3

S Band Radar
PRF1 = 800Hz Blind Sped = 80 knots

Doppler Filters at PRF2 PRF1 = 500Hz Blind Speed = 50 knots

1 2 3 4 5 1 2 3 4 5 1 2 3 4 5 1 2 3 4

Doppler velocity

Each filter 100 Hz wide (10 knots) C1 = b1 m2 = 1 modulo m1 C2 = b2 m1 = 1 modulo m2 C1 = b1 m2 = 1 modulo m1 C1 = b1 5 = 1 modulo 8 b1 = 5 C2 = b2 8 = 1 modulo 5 b2 = 2
IEEE New Hampshire Section IEEE AES Society

A1 = 5 A2 = 3

m1 = 8 m2 = 5

Rc = (C1A1 + C2A2) modulo (40) = [5(5x5) + 3(8x2)] modulo (40) = (125+48) modulo (40) = 173 modulo (40) = 13 True velocity = 130 knots
Radar Systems Course 42 Pulse Doppler 11/1/2009

Example - Chinese Remainder Theorem

Shoe Length of 4 Mens Feet Bob m1 = 7 inches Larry m2 = 8 inches Moe m3 = 9 inches Curly m4 = 11 inches Bob Larry Moe Curly Measure of a Room (Remainder) 2 inches remainder = A1 5 inches remainder = A2 5 inches remainder = A3 6 inches remainder = A4

WHAT IS THE LENGTH OF THE ROOM ?? L = ( C1A1 + C2A2 + C3A3 + C4A4 ) modulo (m1 m2 m3 m4 ) m1 m2 m3 m4 = 5544 C1 = b1 x m2 m3 m4 = 1 modulo m1 b1 x 8 x 9 x 11 = 1 modulo 7 b1 x (7+1) x (7+2) x (7+4) = 1 modulo 7 8 b1 = 1 modulo 7 b1 = 1 L = [A1(792x1) + A2(693x5) + A3(616x7) + A4(504x5)] modulo 5544 = [2(792) + 5(3465) + 5(4312) + 6(2520)] modulo 5544 = [1584 + 17,325 + 21,560 + 15,120] modulo 5544 = 149 inches
Radar Systems Course 43 Pulse Doppler 11/1/2009

IEEE New Hampshire Section IEEE AES Society

Outline

Introduction Problem perspective
Burst Waveforms and their properties The impact of Moores Law on radar Signal Processing
Past, present, and the future

Pulse Doppler Processing Techniques

Description of pulse Doppler processing Low PRF Example Moving Target Detector (MTD) Range and Doppler Ambiguities Ambiguity Resolution - Chinese remainder theorem The Ambiguity Function Preview of Airborne Pulse Doppler issues

Radar Systems Course 44 Pulse Doppler 11/1/2009

Summary
IEEE New Hampshire Section IEEE AES Society

Quick Matched Filter Review

Matched Filter is the cross correlation between :

Received signal (plus noise) , and A replica of the transmitted signal Matched Filter Output =

s( t ) = u ( t ) e 2 j f T t
TR = Round trip time
delay to target

s R (t ) s (t TR )dt

For low S/N assumed:

Autocorrelation of transmitted signal It is assumed that Doppler velocity of target is zero

Usually the target is moving and the Doppler frequency of the target is not zero Then, the output of matched filter is the cross correlation of the transmitted signal and the received Doppler shifted echo.

Radar Systems Course 45 Pulse Doppler 11/1/2009

IEEE New Hampshire Section IEEE AES Society

The Ambiguity Function

The Ambiguity Function is the squared magnitude of the cross correlation (output of matched filter) of the transmitted signal and the received Doppler shifted echo. Studying (analytically and graphically) the two dimensional properties of the Ambiguity Function as both :
Time delay (range), and Doppler frequency ( Doppler velocity)

are varied, can give great insight into understanding many of the waveforms properties, in particular:
Target resolution, Waveform measurement accuracy, Response to various types of clutter, and Ambiguities in Doppler velocity and range

Radar Systems Course 46 Pulse Doppler 11/1/2009

IEEE New Hampshire Section IEEE AES Society

The Ambiguity Function

The Ambiguity Function is the squared magnitude of the cross correlation (output of matched filter) of the transmitted signal and the received Doppler shifted echo. Thus, with some algebraic manipulation *

(TR , f D ) =

u(t ) u (t + TR ) e 2 j f D t dt
2

Thus, the ambiguity function is (TR , f D ) TR is the round trip time delay to the target f D is the Doppler shift of the target
2 j fT t and s(t ) = u( t ) e

* See Skolnik Reference 1, pp 329-330 for details IEEE New Hampshire Section IEEE AES Society

Radar Systems Course 47 Pulse Doppler 11/1/2009

Properties of the Ambiguity Function

Maximum value of the ambiguity function = (2 E )2 At true location of target TD = 0 When, f D = 0 Note: s(t ) = u(t )e 2 j f0 t Total volume under surface of ambiguity function = (2 E )2 Behavior along TR axis (TR ,0) = u(t ) u (t + TR ) dt Square of autocorrelation function of u(t )
2 2 2 2

Behavior along frequency, f D , axis (0, f D ) = u (t ) e Square of inverse Fourier Transform of u 2 (t )

2 j f0 t

dt

A good model of the ambiguity function, suggested by Skolnik, is a box of sand 2 Total volume of sand is = (2 E ) , The sand may be in different 2 piles, but its volume is constrained to be = (2 E )
IEEE New Hampshire Section IEEE AES Society

Radar Systems Course 48 Pulse Doppler 11/1/2009

Three General Classes of Ambiguity Functions

Typically, the magnitude, not the magnitude squared is plotted

(TR , f D )

fD

TR
Knife Edge (ridge)

(TR , f D )

(TR , f D )

fD

fD

TR
Thumbtack
Radar Systems Course 49 Pulse Doppler 11/1/2009

TR
Bed of Spikes
IEEE New Hampshire Section IEEE AES Society

Three General Classes of Ambiguity Functions

Linear FM pulse, where range and Doppler are coupled

(TR , f D )

f D Knife Edge (ridge) Used to measure one parameter:

TR
Knife Edge (ridge)

range , Doppler, or a linear combination of range and Doppler Examples : a single rectangular pulsed sine wave or a single rectangular linear FM pulse

(TR , f D )

(TR , f D )

fD

fD

TR
Thumbtack
Radar Systems Course 50 Pulse Doppler 11/1/2009

TR
Bed of Spikes
IEEE New Hampshire Section IEEE AES Society

Three General Classes of Ambiguity Functions

(TR , f D )

fD

TR
Knife Edge (ridge)

Bed of Spikes Used to measure both range , Doppler with ambiguities Example : a burst of N pulses of sine wave

(TR , f D )

(TR , f D )

fD

fD

TR
Thumbtack
Radar Systems Course 51 Pulse Doppler 11/1/2009

TR
Bed of Spikes
IEEE New Hampshire Section IEEE AES Society

(TR , f D )

Three General Classes of Ambiguity Functions

fD

Thumbtack Examples : pseudorandom noise waveforms (rarely used in radar)

TR
Knife Edge (ridge)

(TR , f D )

(TR , f D )

fD

fD

TR
Thumbtack
Radar Systems Course 52 Pulse Doppler 11/1/2009

TR
Bed of Spikes
IEEE New Hampshire Section IEEE AES Society

Ambiguity Function Rectangular Pulse

Ambiguity Function for two simple single sine wave pulses, each with different pulse widths Examples - 2D slices across Ambiguity Function fD f
D

1/

TR

1/

TR

Long pulsewidth

Short pulsewidth

f D = Doppler frequency shift

Radar Systems Course 53 Pulse Doppler 11/1/2009

TR = Time delay

= pulsewidth
IEEE New Hampshire Section IEEE AES Society

Ambiguity Function of Rectangular Pulse

= pulsewidth

(TR , f D )

fD

1/

TR

2/
3/
Triangular shape along time axis

(sin x ) / x shape along frequency axis

Radar Systems Course 54 Pulse Doppler 11/1/2009

Adapted from Rihaczek, in Skolnik, Reference 13

IEEE New Hampshire Section IEEE AES Society

Ambiguity Function of Linear FM Pulse

(TR , f D )

fD

TR

Adapted from Rihaczek, in Skolnik, Reference 13

Radar Systems Course 55 Pulse Doppler 11/1/2009

IEEE New Hampshire Section IEEE AES Society

Ambiguity Function of Linear FM Pulse

fD

f D = kTR

Ridge (knife edge) in Ambiguity diagram illustrates range Doppler coupling in linear FM waveform In this case, BT >> 1 Angle of ridge is determined by the slope B / T

B 1/ T T
T

TR

1/ B

T = Pulsewidth B = Bandwidth = f 2 f1
IEEE New Hampshire Section IEEE AES Society

Radar Systems Course 56 Pulse Doppler 11/1/2009

Ambiguity Function for a Burst of Five Rectangular Pulses

Bed of Spikes
fD
Spikes Illustrate: Blind Speeds & Blind ranges

1 / TP

TR

TD TP

= total duration = pulse repetition period = pulse duration

TP

Pulse radars usually use a filter matched to one pulse and then integrate the N pulses 5 pulse burst

TP
Radar Systems Course 57 Pulse Doppler 11/1/2009

TD
IEEE New Hampshire Section IEEE AES Society

Ambiguity Diagram for Phase Coded Pulse

13 Bit Barker Code
(TR , f D )

fD

TR

Adapted from Rihaczek, in Skolnik, Reference 13

Radar Systems Course 58 Pulse Doppler 11/1/2009

IEEE New Hampshire Section IEEE AES Society

Outline

Introduction Problem perspective
Burst Waveforms and their properties The impact of Moores Law on radar Signal Processing
Past, present, and the future

Pulse Doppler Processing Techniques

Description of pulse Doppler processing Low PRF Example Moving Target Detector (MTD) Range and Doppler Ambiguities Ambiguity Resolution - Chinese remainder theorem The Ambiguity Function Preview of Airborne Pulse Doppler issues

Radar Systems Course 59 Pulse Doppler 11/1/2009

Summary
IEEE New Hampshire Section IEEE AES Society

Pulse Doppler Radar Techniques on Airborne Platforms

F-16 APG-66 , 68

Courtesy of US Air Force

E-2C APS-125
Courtesy of US Navy

AWACS E-3A APY-1

Courtesy of US Air Force

Radar Systems Course 60 Pulse Doppler 11/1/2009

IEEE New Hampshire Section IEEE AES Society

Doppler Velocity - Range Ambiguity Issues

Unambiguous Range (nmi)
400 3000 100
Hz Hz G

40

10

Combining

First Blind Speed (knots)

1000

300

100

d 5 an 43 B Hz F G nd 3 VH Hz Ba 1. G d 2 HF U an 3. B d L Hz an G B .4 S 9 d n Ba X

0 22

f PRF VB = 2
B d an 35 Hz G

and

Ka

c RU = 2 f PRF
Yields

30

10 0.1

10

100

c VB = 4 RU

Pulse Repetition Rate (KHz) Example R U = 10 nmi f PRF ~ 8 KHz VB ~ 270 knots X-Band Fighter Radar
Radar Systems Course 61 Pulse Doppler 11/1/2009

IEEE New Hampshire Section IEEE AES Society

Airborne Radar Clutter Characteristics

Illustrative example Without Pulse-Doppler ambiguities

Doppler frequency of mainbeam clutter depends on scan direction Doppler frequency of target depends on scan direction and aspect angle
Viewgraph Courtesy of MIT Lincoln Laboratory Used with permission

Radar Systems Course 62 Pulse Doppler 11/1/2009

IEEE New Hampshire Section IEEE AES Society

Summary

Pulse Doppler techniques can be used to optimally reject various forms of radar clutter Moving Target Detector is an example of near-optimum Doppler processing and associated adaptive thresholding techniques implemented in low PRF radars Ambiguities in range and Doppler velocity can be resolved by transmitting multiple bursts of pulses with different PRFs The Chinese remainder Theorem is a useful tool in resolving these ambiguities The ambiguity function is a useful tool to understand the time and frequency properties of different waveforms

Radar Systems Course 63 Pulse Doppler 11/1/2009

IEEE New Hampshire Section IEEE AES Society

Homework Problems

From Skolnik (Reference 1)

Problems 3-9, 3-10, 3-11, 3-12, 3-13, 3-14 and 3-15

Radar Systems Course 64 Pulse Doppler 11/1/2009

IEEE New Hampshire Section IEEE AES Society

References
1. Skolnik, M., Introduction to Radar Systems, McGraw-Hill, New York, 3rd Ed., 2001 2. Barton, D. K., Modern Radar System Analysis, Norwood, Mass., Artech House, 1988 3. Skolnik, M., Editor in Chief, Radar Handbook, New York, McGraw-Hill, 3rd Ed., 2008 4. Skolnik, M., Editor in Chief, Radar Handbook, New York, McGraw-Hill, 2nd Ed., 1990 5. Nathanson, F. E., Radar Design Principles, New York, McGraw-Hill, 1st Ed., 1969 6. Richards, M., Fundamentals of Radar Signal Processing, McGraw-Hill, New York, 2005 7. Schleher, D. C., MTI and Pulsed Doppler Radar, Artech, Boston, 1991 8. ODonnell, R. M. and Cartledge, L., Description and Performance Evaluation of the Moving Target Detector, Project Report ATC-69, MIT Lincoln Laboratory, 1977
IEEE New Hampshire Section IEEE AES Society

Radar Systems Course 65 Pulse Doppler 11/1/2009

References (continued)
9. Rohling, H., Doppler Processing, Waveform Design and Performance Measures for Some Doppler and MTD Radars, Ortung und Navigation, March 1988 and January 1982 10. Bassford, R. et al, Test and Evaluation of the Moving Target Detector (MTD) Radar, FAA Report, FAA-RD-77-118, 1977 11. Rabinowitz, S. J., et. al, Applications of Digital Technology to Radar, Proceedings of the IEEE, Vol. 73, No 2, pp 325-339 12 Karp, D. Moving Target Detector Mod II Summary Report, Project Report ATC 96, MIT Lincoln Laboratory. 1981 13. Skolnik, M., Editor in Chief, Radar Handbook, New York, McGrawHill, 1st Ed., 1970 14. Delong, D. F. and Hofffsteter, E., On the Design of Optimum Waveforms for Clutter Rejection, IEEE Information Theory, Vol IT13, no. 3, July 1967 15. Taylor, J. W. et. al., Design of a New Airport Surveillance Radar ASR-9, Proceedings of the IEEE, Vol. 73, No 2, pp 284-289

Radar Systems Course 66 Pulse Doppler 11/1/2009

IEEE New Hampshire Section IEEE AES Society

Acknowledgements

Mr. C. E. Muehe Dr James Ward

Radar Systems Course 67 Pulse Doppler 11/1/2009

IEEE New Hampshire Section IEEE AES Society