Hardware in the Loop Radar Clutter
Simulation
Presenter: Jurgen Strydom
Systems Engineer & Signal Analyst
Experimental EW Systems, CSIR
Email: [email protected]
Co-authors: Jacques Cilliers, Andre McDonald, Klasie Olivier
7 November 2012
Outline
• Hardware in the Loop simulation
• Overview of radar system specifications
• Radar testing and evaluation
• Digital Radio Frequency memory
• Radar environment simulation
– Complex targets
– ECM
– Clutter
• Radar clutter simulation
– Ground clutter
– Sea clutter
– Sidelobe clutter for an airborne platform
• Conclusion
© CSIR 2012 Slide 2
Hardware in the Loop Simulation
• Hardware-in-the-loop simulation is a well established technique used in the
design and evaluation of hardware systems
• Traditional testing of systems relied solely on field trails
• Hardware in the loop replaces the actual environment with a simulated
environment
• The environment is simulated on a hardware platform, and connected to the
system under test
• Radar systems are connected by RF either through air coupling or by cable
© CSIR 2012 Slide 3
Radar Environment
© CSIR 2012 Slide 4
Radar Types
© CSIR 2012 Slide 5
• Radars by Function
• Weather Avoidance
• Navigation & Tracking
• Search & Surveillance
• High resolution Imaging (mapping)
• Proximity Fuses
• Countermeasures
• Examples
• Search radar (High power, pulsed, low PRF, long pulse
lengths, lower frequency bands (L/S), low resolution)
• Airborne radar (Low power, pulsed, medium to high PRF,
higher frequency bands (X), high resolution)
Radar waveforms
© CSIR 2012 Slide 6
• Continuous wave (CW)
• Frequency modulated continuous wave (FMCW)
• Pulsed
• Non-coherent
• Coherent
• Low PRF (1 - 3 KHz)
• Medium PRF (10 - 30 KHz)
• High PRF (100 - 300 KHz)
• Spread spectrum - Low Probability of Intercept (LPI)
• Multi mode (track while scan)
• Examples
• Missile fuse (FMCW)
• SAR / ISAR (pulsed Low / Medium PRF)
Environmental Factors
© CSIR 2012 Slide 7
• Influence on radar performance
• Target Glint
• Multipath
• Signal propagation effects
• Noise
• Clutter
• Sea clutter
• Ground clutter (trees, grass, buildings, mountains)
• Countermeasure clutter (e.g. Chaff)
• Airborne clutter (birds, clouds, rain/snow, small aircraft)
• Intentional / Unintentional Jamming / System Blanking
• Line of sight
• Commercial Spectrum usage (cell-phones, etc., etc.)
Radar testing and evaluation
• Radar testing and evaluation is becoming difficult for modern radars
because of their adaptive nature
• Two possible approaches:
• Development teams can design and build their own test equipment
for each radar
• Generic test equipment can be designed to test a class of radars
• The second approach is well suited to:
• Organizations such as defence evaluation and research institutes
(DERI)
• Agencies that specialize in independent review acceptance testing
and optimisation of operational utilisation
© CSIR 2012 Slide 8
Radar testing and evaluation
• Hardware in the loop testing allows for the evaluation of critical
functionality, before optimisation of weight and size
• Reduces number of measurement field trails, as well as cost of
developing system
• Continuous testing of repeatable scenarios during the development
of the radar
© CSIR 2012 Slide 9
CSIR DRFM
The DRFM kernel
• Captures radar transmit pulse
with ADC
• FPGA stores data in digital
memory
• FPGA reads data at required
time delay
• Radar pulse transmitted through
DAC
• Local oscillator used in both RF
up and down conversion chain
to guarantee coherency
• This architecture is commonly
referred to as a DRFM
© CSIR 2012 Slide 11
The HIL simulator
• Captures radar transmit pulse
with ADC
• FPGA stores data in digital
memory and a modulation is
applied
• FPGA reads data at required
time delay
• Radar pulse transmitted through
DAC
• Local oscillator used in both RF
up and down conversion chain
to guarantee coherency
• This architecture is commonly
referred to as a Radar
Environment simulator
© CSIR 2012 Slide 12
The HIL simulator
• DRFM and Radar environment simulator processes match
• DRFM is a modular building block
• Different firmware on DRFM allows for different types of modulations
to signals which represents different radar environmental elements
• Complex targets
• ECM techniques
• Radar clutter
• The combination of each of these DRFM channels results in the
Hardware in the loop radar environment simulator
© CSIR 2012 Slide 13
Radar Environment Simulation
Advances in DRFM technology allows for the real-time simulation of a
wide range of targets, ECM techniques, and environmental effects
Radar Environment Simulation
Target Simulation Electronic Attack Target Environment
Point targets
Multi scatterer targets
Jamming Clutter
© CSIR 2012 Slide 14
Complex Targets
• A conventional DRFM simulates a target by simply re-transmitting the radar
pulse
• Approximate target as a single scatterer
• Assumption is made that radar target is only present in a single range cell of
the radar
© CSIR 2012 Slide 15
Complex Targets
• Today's radars use High Range
Resolution (HRR) profile techniques
• Separation of scattering points on a
single target
• This creates a "complex" target return
• A conventional DRFM simulates a target by simply re-transmitting the radar
pulse
• Approximate target as a single scatterer
• Assumption is made that radar target is only present in a single range cell of
the radar
• Can be used for Non-cooperative target recognition (NCTR)
© CSIR 2012 Slide 16
Complex Targets
• High Range Resolution target profile [1]
[1] J.C. Smit, J.E. Cilliers, E.H. Burger, "Comparison of MLFMM, PO and SBR
for RCS Investigations in Radar Applications," IET radar conference 2012.
© CSIR 2012 Slide 17
ECM
• Deception techniques to interfere with target detection and tracking
• Non-coherent jamming (injection of high powered noise)
• Manipulation of transmitted radar pulse
– Range
– Doppler
– Amplitude information
• Goal is to have radar interpret false target as actual target
• Common ECM techniques for tracking radar
– Range gate pull off
– Velocity gate pull off
© CSIR 2012 Slide 18
Radar Clutter
© CSIR 2012 Slide 19
Clutter from literature
© CSIR 2012 Slide 20
Considerations for high fidelity clutter generation
• Clutter Radar Cross Section (RCS)
• Number of discrete scatterers
• Spatial extent of clutter
• Velocity extent of clutter (Doppler spectrum)
• Wavelength dependence
• Amplitude distribution
• Spatial correlation
• Polarization properties
From [4]
Radar Clutter from literature
© CSIR 2012 Slide 21
From [1]
© CSIR 2012 Slide 22
Amplitude distribution for radar clutter
Grasslands measured with HH polarisation at X band with an incidence
angle of 20 degrees
Radar Clutter from literature
From [2]
Radar Clutter from literature
© CSIR 2012 Slide 23
10dB attenuation
30dB attenuation
50dB attenuation
Clutter returns can be very large:
From [3]
Radar Clutter from literature
© CSIR 2012 Slide 24
Clutter becomes more spiky as range resolution increases
From [3]
Radar Clutter from literature
© CSIR 2012 Slide 25
Spatial behaviour of clutter
No real correlation in spatial dimension, but scenario depended
From [1]
Radar Clutter from literature
© CSIR 2012 Slide 26
Temporal behaviour of clutter
Temporal correlation can be large
From [1]
The Problem
© CSIR 2012 Slide 27
• Clutter returns interfere with the object of interest (target)
• The performance in clutter is a critical aspect of the radar
• Real world testing of a radar against all types of clutter for all
possible types of scenarios is costly and difficult to repeat
• Software simulation cannot take all the finer details of the
complete design and implemented system into account
• Severely limited with software simulation if you are required to
verify a radar purchased from a 3rd party
The Solution
© CSIR 2012 Slide 28
• Hardware in the loop simulation on DRFM based hardware
• Statistical modelling of clutter, NOT recorded data
• Playback of recorded data is radar and configuration dependent
Synthetic Clutter Simulation
© CSIR 2012 Slide 29
Goal:
Simulate clutter in real time as realistic as possible
Reality:
Real time is pretty fast, and realistic is computationally complex, so these
requirements have to be interpreted somewhat...
Realistic land clutter varies drastically depending on the terrain
Land clutter includes:
Soil, rocks, trees, grass, shrubs, short vegetation, road surfaces, urban
areas, dry snow, wet snow, etc.
A single range line can contain any combination of these
Synthetic Clutter Simulation
© CSIR 2011 Slide 30
Synthetic Clutter Simulation
• Wind direction does not change relative to radar because of the
constant look angle
• Statistics remain constant
• Range line can be divided up into range segments to re-create the
change in properties of the illuminated area with range
Synthetic Clutter Simulation
© CSIR 2011 Slide 32
• Segments contain an arbitrary number of time correlated scatter points
• Each segment is set independently of the other segments
• Each segment has is own statistics
• Amplitude
• Spectrum
• Probability Density Function (PDF)
• All segments combine to produce a single range line
Synthetic Clutter Simulation
© CSIR 2012 Slide 33
Record accurate radar
data
Figure on left:
Data captured with the
measurement radar of
the CSIR
Synthetic Clutter Simulation
© CSIR 2012 Slide 34
Synthetic Clutter Simulation
© CSIR 2012 Slide 35
What has been achieved thus far:
• Correlated ground clutter (Rayleigh, Weibull, Log-Normal)
• Mainbeam for stationary platform and moving platform
• Gaussian approximation to clutter bandwidth
• 2 million+ independent clutter scatterers in a range line
• 500 MHz instantaneous bandwidth
• Input pulse lengths from 50 ns up to 300 us
• PRF from 0.8 kHz to 300 kHz
• Synthetic Clutter Simulation System covers large number of radar
systems
Sea Clutter Simulation
Current research:
• Clutter simulation for a seaborne search radar
The Challenge:
• Sea clutter statistics not straightforward to connect to sea
environment, there are no good solutions in literature as of yet
• Clutter statistics are dependent on many variables (wind direction
and magnitude, wave direction and magnitude, water depth, etc.)
• Many different scattering mechanisms, complex models
© CSIR 2012 Slide 36
Sea Clutter Simulation
• Scattering mechanisms for sea clutter have large and small scale
• Spiky nature of clutter caused by the small scale scattering
mechanisms: ripple, spray and foam.
• Large scale (swell) decorrelates slowly, small scale decorrelates quickly
• Breaking waves
© CSIR 2012 Slide 37
Sea Clutter Simulation
Compound model for sea clutter:
• Texture (tau): non-negative random process; takes into account the
local mean power (large scale)
• Speckle (x): complex Gaussian process, takes into account the local
backscattering (small scale)
• For a Gamma texture the K distribution results (amplitude PDF):
© CSIR 2012 Slide 38
Synthetic Clutter Simulation
• Sufficient for low resolution radars
and for large grazing angles
• Many scattering points in a range
cell cause Gaussian statistics
• Magnitude of a Gaussian signal is
Rayleigh
• K distribution represents sea
clutter statistics
• Sea clutter is more spiky
• Fewer scattering points in a
range cell causes more spikes
0 100 200 300 400 500 600 700 800 900 10000
1
2
3
4
5
6
7
8
0 100 200 300 400 500 600 700 800 900 10000
1
2
3
4
5
6
7
8
Range
Range
Am
plit
ud
e
Am
plit
ud
e
Rayleigh distributed clutter
K distributed clutter
© CSIR 2012 Slide 39
Sea Clutter Measurements
© CSIR 2012 Slide 40
Sea Clutter Measurements
© CSIR 2012 Slide 41
Sea Clutter Measurements [2]
Low sea state High sea state
Rayleigh
Incre
asin
g S
pik
yn
ess (
K-D
ist)
A. McDonald, J.E. Cilliers, "Autoregressive-to-anything process model of
maritime clutter and targets," IET radar conference 2012.
© CSIR 2012 Slide 42
Sea Clutter Measurements
• On-line database freely available of small boats in sea clutter
– Radar data database of this quality and size are very rare
– Very well instrumented targets in sea clutter
– Database is well documented in a user guide
– Aim is small boat detection in sea clutter
– Over 100 users from over 16 countries
– Did I mention it is free?
• Available at:
http://www.csir.co.za/small_boat_detection/
Synthetic Clutter Simulation
• Wind and wave direction changes relative to rotation angle of radar
• Therefore statistics change
• Segments & Sectors are used to re-create a scene
• Time multiplexing can be used to create a 360 degree scenario
Synthetic Clutter Simulation
Rotation angle
Am
plit
ud
e
© CSIR 2012 Slide 45
• Time multiplexing can be used to create a 360 degree scenario
• Scenario divided into sectors based on desired statistics in that
direction
– Divisions based on regions of similar statistical properties
– Divided in a statistically meaningful way
– For example :1 sector of 80 degrees for an area looking at a
mountain and 35 sectors of 8 degrees to capture the sea surface
with the relative wind and wave direction changes
Synthetic Clutter Simulation
© CSIR 2012 Slide 46
Synthetic Clutter Simulation
© CSIR 2012 Slide 47
Synthetic Clutter Simulation
© CSIR 2012 Slide 48
Current research:
• Clutter for a moving airborne platform
The Challenge:
• For stationary radar platforms the mainlobe is sufficient
• For moving airborne radar platforms the mainlobe is not the only
contributing factor to the radar range Doppler map
• Antenna sidelobes of the radar becomes a large contributing factor
Synthetic Clutter Simulation
© CSIR 2012 Slide 49
From: N. Levanon, Radar Principles
Mainlobe only scenario
Synthetic Clutter Simulation
© CSIR 2012 Slide 50
From: G. Morris and L. Harkness, Airborne Pulsed Doppler Radar
Synthetic Clutter Simulation
© CSIR 2012 Slide 51
Data from:
Recorded Data Airborne Range Doppler map
Synthetic Clutter Simulation
© CSIR 2012 Slide 52
Building the airborne radar Range Doppler map:
• It is relatively easy to adapt the mainbeam of the stationary ground
based platform to the moving airborne platform case
• Apply a Doppler offset that corresponds to the look direction and
range of the intersection with the surface
• Airborne platform (because of its movement) spends less time
looking at the same patch of ground, thus less correlation between
pulses, which results in a wider bandwidth
Synthetic Clutter Simulation
© CSIR 2012 Slide 53
Spectrum of mainbeam Spectrum of sidelobes
- Gaussian Doppler power
spectrum
- Simple to implement
- Jakes Doppler power spectrum
- Computationally expensive to
implement
Synthetic Clutter Simulation
© CSIR 2012 Slide 54
Range (k
m)
Doppler velocity (m/s)
Sim
ula
ted a
irborn
e ra
dar ra
nge D
opple
r map
00.1
0.2
0.3
0.4
0.5
0.6
0.7
0.8
0.9
1-2
50
-200
-150
-100
-50 0
50
100
150
200
250
0 0.5 1 1.5 2 2.5
x 104
-10
-5
0
5
10
15
20
Doppler Frequency
Magnitude
Sidelobe simulation
Simulated range Doppler map with the DRFM sidelobe clutter algorithm
Synthetic Clutter Simulation
Challenges:
• Difficult to split a single scenario over multiple DRFMs
• "High" latency and "low" transfer rate between seperate DRFM
systems (relative to on chip)
• This bottleneck makes it difficult to sync clutter scatterers
Possible Solution:
• Mainlobe / Sidelobe is an ideal split for the use of multiple DRFMs
• Statistics of mainlobe is different and uncorrelated to that of the
sidelobe
• Thus one can get away with minimum sync of scatterers
© CSIR 2012 Slide 55
Synthetic Clutter Simulation
Technological tradeoffs:
• DRFM is processing power limited
• Bandwidth spectral shaping quality and accuracy
• Fidelity of clutter (number / update frequency, of clutter samples )
• Complexity of statistical distribution shape
• Rayleigh (least complex)
• Weibull (medium complexity)
• Log-Normal (medium complexity)
• K-Distribution (most complex)
• Trade-off between a high fidelity model over a small area or a low
fidelity model over a large area
© CSIR 2012 Slide 56
Future Synthetic Clutter Simulation
© CSIR 2012 Slide 57
Sea Clutter Airborne Radar Ground Clutter Weather Clutter
K-Distributed PDF
Rayleigh PDF
Log-Normal & Weibull PDF
Mainbeam clutter
Sidelobe clutter
Clouds / Rain / Hail / Snow
Exponential Doppler Spectrum
Moving airborne platform
Stationary platform
Gaussian Doppler Spectrum
Moving platform
Completed:
In-Progress:
Future research: Arbitrary PDF
Counter Measure Clutter
Chaff
Multipath
Integrated capability
Scenario simulation & control
RCS / HRR prediction HIL simulation
System under test
Radar under test
International Conferences
• Past conferences (2010 / 2011)
• Conferences for 2012
The End
© CSIR 2012 Slide 60
References
[1] J.B. Billingsley, "Low-Angle Radar Land Clutter,
Measurements and Empirical Models"
[2] F.T. Ulaby and M.C. Dobson, "Radar Scattering Statistics
for Terrain"
[3] F.E. Nathanson, Radar Design Principles, Signal
Processing and the Environment"
[4] D.K. Barton, "Modern Radar System Analysis"
[5] K.D. Ward, R.J.A. Tough and S. Watts, "Sea Clutter:
Scattering, the K Distribution and Radar Performance"
The End
© CSIR 2012 Slide 61
Special thanks:
Airborne radar data provided by:
Contact information:
Jurgen Strydom