The NRL Multi Aperture SAR (NRL MSAR): System Description and Recent - - PowerPoint PPT Presentation

β–Ά
the nrl multi aperture sar nrl msar
SMART_READER_LITE
LIVE PREVIEW

The NRL Multi Aperture SAR (NRL MSAR): System Description and Recent - - PowerPoint PPT Presentation

Jul 14, 2023 443 likes β€’817 views

UNCLASSIFIED CLASSIFICATION The NRL Multi Aperture SAR (NRL MSAR): System Description and Recent Results Luke Rosenberg Defence Science and Technology Organisation, Australia Mark Sletten, Naval Research Laboratory, USA 1 UNCLASSIFIED

slide-1
SLIDE 1

UNCLASSIFIED

1

CLASSIFICATION

The NRL Multi Aperture SAR (NRL MSAR): System Description and Recent Results

Luke Rosenberg Defence Science and Technology Organisation, Australia Mark Sletten, Naval Research Laboratory, USA

slide-2
SLIDE 2

UNCLASSIFIED

2

  • Motion in SAR imagery
  • Single vs. Multi-aperture SAR
  • The Velocity SAR algorithm for focussing moving scatterers
  • Demonstration of the VSAR algorithm using the NRL FOPAIR
  • Initial results from the Airborne MSAR system
  • Enhanced VSAR
  • Future plans

Acknowledgements:

  • Naval Research Laboratory, Remote Sensing Division:

Mark Sletten, Steve Menk, Jakov Toporkov, Bob Jansen

  • Naval Research Laboratory, Radar Division:

Raghu Raj, Denny Baden

Outline

slide-3
SLIDE 3

UNCLASSIFIED

3

Effects of Scene/Target Motion on SAR Signatures

  • Relative motion between platform and scene is fundamental to SAR
  • SAR processors assume scene is stationary: Scene motion causes distortion
  • Constant radial motion: azimuthal offsets, a.k.a. β€œtrain off the track” distortion
  • Radial acceleration and azimuthal motion: azimuth defocusing
  • Issue is significant for marine applications, since complex motion is pervasive
  • Signatures not only displaced, but smeared as well

Real Aperture Radar Image SAR Image (emulated)

NRL FOPAIR Imagery, Small boat on the Chesapeake Bay

slide-4
SLIDE 4

UNCLASSIFIED

4

Formation of a Standard SAR Image

1 phase center

Time Space

Synthetic aperture Road Cars Image

Azimuth

slide-5
SLIDE 5

UNCLASSIFIED

5

Formation of an MSAR Image Stack

M phase centers

Time Space

Road Cars Image

Azimuth

slide-6
SLIDE 6

UNCLASSIFIED

6

Formation of an MSAR Image Stack

M phase centers

Time Space

Image Time Stack

Azimuth

𝑒 = 𝑒0 𝑒 = 𝑒0 + βˆ†π‘’ 𝑒 = 𝑒0 + 2βˆ†π‘’ 𝑒 = 𝑒0 + 3βˆ†π‘’ 𝑒 = 𝑒0 + (𝑁 βˆ’ 1)βˆ†π‘’ Road Cars

  • Images look the same: motion information is in the phase of the complex pixels
  • Images look the same: motion information is in the phase of the complex pixels
slide-7
SLIDE 7

UNCLASSIFIED

7

Velocity SAR (VSAR) Processing

  • Doppler processing converts the image time-stack into a velocity stack
  • Shifting each velocity image by

𝑆 π‘Š

π‘ž π‘€πΈπ‘π‘ž corrects azimuthal misplacement

  • An incoherent sum down the corrected velocity stack forms a single corrected image

Time Stack Velocity Stack Shifted Velocity Stack

FFT

Azimuth Azimuth Azimuth

Doppler Frequency/velocity

Corrected Image (Incoherent sum) Azimuth

slide-8
SLIDE 8

UNCLASSIFIED

8

NRL FOPAIR

  • NRL Focused Phased Array Imaging Radar (NRL FOPAIR)
  • Updated version of UMass FOPAIR (McIntosh and Frasier, 1995)
  • Mimics a SAR: Receive array elements rapidly and sequentially scanned
  • Generates image time-stacks with a high frame rate (780 fps β€œmovies”)
  • X-band (9.85 GHz) fully polarimetric, 200 MHz BW (0.75 m resolution)
  • 16-module receive array easy to reconfigure
slide-9
SLIDE 9

UNCLASSIFIED

9

M Apertures deff

FOPAIR as an MSAR Test Bed

Time Space

MSAR FOPAIR

Time Space

…

slide-10
SLIDE 10

UNCLASSIFIED

10

SAR Image (emulated)

First Reported Demonstration of VSAR-Based Signature Correction

  • NRL FOPAIR imagery of a small boat used to demonstrate VSAR signature correction

Sletten, IEEE Trans. Geoscience Remote Sens., Vol. 51, No. 5, May 2013

VSAR Image (emulated)

slide-11
SLIDE 11

UNCLASSIFIED

11

  • X-band (9.875 GHz CF)
  • Bandwidth: 220 MHz
  • Waveform: LFM, both up and down chirps
  • Peak and average power: 1.4 kW, 210 W
  • Phase centers: 32 along-track
  • Polarization: VV
  • Platform: Saab 340
  • IMU: Novatel
  • Data recorder: NRL custom-built, 4-channel, 800 MB/s sustained

~ 1200 m ~ 1.5 km

22Β° 45Β°

NRL MSAR Basic Specifications

slide-12
SLIDE 12

UNCLASSIFIED

12

Saab 340 Tx Down-chirp Tx Up-chirp Rx 1-16 Novatel IMU (behind Rx modules)

NRL MSAR Aircraft and Radome

slide-13
SLIDE 13

UNCLASSIFIED

13

  • Use two transmit antennas to double number of phase centers
  • Minimum and maximum unambiguous velocities, assuming VSAR-type processing:

At Vp=70 m/s (Saab 340)

  • Cycle through all 32 combinations of Tx and Rx antennas in 320 microsec (8 pulses)

32 Resulting Phase Centers 2 Transmit Horns 16 Physical Receive Elements deff

deff β‰ˆ d/2 = 5.25 cm

s m d V v

eff p

/ 10 4

max

 ο€½ 

s m M d V v

eff p

/ 7 . 2

min

 ο€½ 

d=10.5 cm

32 Phase Center Array

slide-14
SLIDE 14

UNCLASSIFIED

14

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32

Receive Elements Switches U D

Antenna Switching Schematic

Up chirp transmit antenna Down chirp transmit antenna Data acquisition channel 1 Data acquisition channel 2 Data acquisition channel 3 Data acquisition channel 4

slide-15
SLIDE 15

UNCLASSIFIED

15

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32

Receive Elements Switches U Pulse 1 Up-chirp Receive elements 1, 9, 17, 25 Data acquisition channel 1 Data acquisition channel 2 Data acquisition channel 3 Data acquisition channel 4

slide-16
SLIDE 16

UNCLASSIFIED

16

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32

Receive Elements Switches D Pulse 2 Down-chirp Receive elements 1, 9, 17, 25 Data acquisition channel 1 Data acquisition channel 2 Data acquisition channel 3 Data acquisition channel 4

slide-17
SLIDE 17

UNCLASSIFIED

17

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32

Receive Elements Switches U Pulse 3 Up-chirp Receive elements 3, 11, 19, 27 Data acquisition channel 1 Data acquisition channel 2 Data acquisition channel 3 Data acquisition channel 4

slide-18
SLIDE 18

UNCLASSIFIED

18

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32

Receive Elements Switches D Pulse 4 Down-chirp Receive elements 3, 11, 19, 27 Data acquisition channel 1 Data acquisition channel 2 Data acquisition channel 3 Data acquisition channel 4

slide-19
SLIDE 19

UNCLASSIFIED

19

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32

Receive Elements Switches U Pulse 5 Up-chirp Receive elements 5, 13, 21, 29 Data acquisition channel 1 Data acquisition channel 2 Data acquisition channel 3 Data acquisition channel 4

slide-20
SLIDE 20

UNCLASSIFIED

20

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32

Receive Elements Switches D Pulse 6 Down-chirp Receive elements 5, 13, 21, 29 Data acquisition channel 1 Data acquisition channel 2 Data acquisition channel 3 Data acquisition channel 4

slide-21
SLIDE 21

UNCLASSIFIED

21

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32

Receive Elements Switches U Pulse 7 Up-chirp Receive elements 7, 15, 23, 31 Data acquisition channel 1 Data acquisition channel 2 Data acquisition channel 3 Data acquisition channel 4

slide-22
SLIDE 22

UNCLASSIFIED

22

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32

Receive Elements Switches D Pulse 8 Down-chirp Receive elements 7, 15, 23, 31 Data acquisition channel 1 Data acquisition channel 2 Data acquisition channel 3 Data acquisition channel 4

slide-23
SLIDE 23

UNCLASSIFIED

23

  • 30 flight hours over the span of 2+ weeks, September 2014.
  • Based out of Newport News VA.
  • After a difficult installation, system worked remarkably well. Some issues:
  • Due to placement of transmit horns, only 28 unique phase centres.
  • Mismatch with up-chirp / down-chirp waveforms - produced low image
  • coherence. Current VSAR results are restricted to 16 phase centres.
  • Two subjects of study
  • Oregon Inlet on the Outer Banks of NC
  • Imaged boats of opportunity, waves, currents, vehicles.
  • Used linear flight patterns (i.e. strip-map).
  • Cooperative vessels in the Southern Chesapeake Bay.
  • Imaged 30 different vessels, both stationary and moving (0-50 kts).
  • Used both linear and circular flight patterns.

Inaugural NRL MSAR Deployment

slide-24
SLIDE 24

UNCLASSIFIED

24

VSAR analysis region

Inaugural NRL MSAR Deployment

Oregon Inlet, NC Outer Banks

slide-25
SLIDE 25

UNCLASSIFIED

25

First VSAR Analysis

Animation: Click to start

Shoaling waves Vehicles

Northbound Southbound

  • VSAR processing significantly reduces smearing of shoaling waves
  • (Faint) vehicle signatures shifted back to bridge
slide-26
SLIDE 26

UNCLASSIFIED

26

Velocity Image Movie

Animation: Click to start Each image shifted by 𝑺

𝑾𝒒 π’˜π‘¬π’‘π’’ to

correct azimuthal displacement

Vehicles

  • Vehicle signatures much more visible than in previous composite image, due to

Doppler filtering inherent in VSAR processing

  • Vehicle speeds projected onto bridge are 64 and 48 mph (speed limit 55 mph)

Northbound Southbound

slide-27
SLIDE 27

UNCLASSIFIED

27

MSAR processing chain

  • Radar collects RAW data – large binary files (~60-160 GB)
  • 1st stage processing (MATLAB):
  • Extracts the 32 phase centres.
  • Baseband conversion and low pass filter.
  • Creates single file for each phase centre (~1-5 GB).
  • 2nd stage processing (C-code / JAVA front end):
  • SAR image formation uses chirp scaling algorithm.
  • Includes range compression and integrated motion compensation.
  • Creates SAR image for each phase centre.
  • 3rd stage processing (MATLAB):
  • Extract small region for processing.
  • Channel balancing.
  • VSAR processing.
slide-28
SLIDE 28

UNCLASSIFIED

28

Adaptive channel balancing

  • VSAR processing assumes SAR magnitude images are identical and there is a reasonable

level of coherency between complex images.

  • Implemented adaptive 2D calibration technique for the SAR images*.
  • Works in image frequency domain.
  • First stage estimates and corrects channel β€˜transfer’ function along each spatial

frequency dimension.

  • Second stage required to balance magnitude in the image domain.
  • Example below shows distribution of the coherence and magnitudes before / after channel

balancing. * Ender, J. H. G. β€˜The airborne Experimental Multi-Channel SAR System AER-II’, European SAR conference, 1996, pp. 49-52.

slide-29
SLIDE 29

UNCLASSIFIED

29

VSAR image improvement

  • Problems identified:
  • 1. Spectral leakage from non-DC velocity images.
  • 2. Velocity components hard to distinguish after non-coherent sum.
  • 3. Loss of dynamic range in VSAR image.
  • Solutions:
  • 1. Identify strong scatterers in the DC velocity image and mask these

pixels in the other velocity images – threshold set as mean of the DC image.

  • 2. Balance the means of the different velocity images relative to the

DC velocity image.

  • 3. Need to mask non-significant scatterers present in each non-DC

velocity image – threshold set as 2-6 std above mean for each image.

  • 4. Further improvements – maximum velocity image and

autoregressive spectral estimate.

slide-30
SLIDE 30

UNCLASSIFIED

31

First VSAR image

  • Original VSAR image with no extra processing.

dB

slide-31
SLIDE 31

UNCLASSIFIED

32

Improved VSAR Image

dB

  • Improved VSAR image with extra processing - filtering removed some details.
slide-32
SLIDE 32

UNCLASSIFIED

33

Maximum VSAR Image

dB

  • Maximum improved VSAR image with extra processing (some extra detail).
slide-33
SLIDE 33

UNCLASSIFIED

34

Improved spectral estimate

dB

  • Final improvement was to introduced a 4th order auto-regressive spectral estimate to

improve velocity resolution / dynamic range.

slide-34
SLIDE 34

UNCLASSIFIED

35

Visualisation – Velocity overlay

  • Find the dominant velocity component in each pixel and overlay it on the VSAR image.
slide-35
SLIDE 35

UNCLASSIFIED

36

Visualisation - 3D Image slice

Breaking wave with 5x5 smoothing window Cars detected on bridge

  • velocities approximately

match speed limit of 55 mph

slide-36
SLIDE 36

UNCLASSIFIED

37

Current / planned work

VSAR

  • Improve coherency across array – utilise 28 unique channels.
  • Investigate VSAR images of charted boats.
  • Develop backprojection code for the circular spotlight mode.
  • Investigate Velocity ISAR algorithm.

Beamforming

  • Modelling of expected performance using the aperture switching scheme.
  • Application of adaptive processing schemes to suppress clutter and detect

targets – i.e. pre / post Doppler STAP. Follow on trial

  • Trial planned for October 2015 focussing on Langmuir Turbulence / small

target detection.

  • Opportunity to test polarimetric MSAR.