US-Australia MOA Meeting October 2005

1
US-Australia MOA Meeting October 2005
Oguz Kazanci and Jeffrey Krolik Duke
UniversityDepartment of Electrical and Computer
EngineeringDurham, NC 27708 Supported by the
CNTPO RD Program
2
Array Beam Pattern with Missing Sensors

• Faulty receive elements can degrade the beam
  pattern significantly depending on their
  locations in the array.
• Beam pattern with 8 missing sensors exhibits 7 dB
  higher peak sidelobes and up to 20 dB higher
  off-angle sidelobes compared to a full array.
• Compensation of missing sensors intended to
  reduce leakage of interferences and
  azimuthally-dependent Doppler-shifted clutter
  into target directed beam.

3
Element-Space Adaptive Channel Compensation (ACC)

• Using good sensors to MMSE interpolate missing
  elements does not explicitly minimize sidelobe
  leakage and thus only indirectly mitigates bad
  channels.
• Idea Adaptively reconstruct full-array snapshot
  such that interference leakage into nominally
  quiet directions is minimized.
• Adaptive channel compensation (ACC) minimizes
  interference leakage into quiet plane-wave
  directions subject to the constraint that the
  data distortion at the good elements is within
  some tolerance.

• Effect of ACC is to make interferers look more
  like plane-waves across the full-array which can
  then be effectively nulled by conventional
  beamformer shading.

4
Doppler-Azimuth Spectrum Element-Apace ACC

• The Doppler-Azimuth plots for Conventional (left)
  and Elementspace ACC from 12 Dec, 2001. Dwell
  171142
• Note the sidelobes of the strong interference in
  the transmitter mainlobe leaking into other
  directions. With element-space ACC, sidelobe
  reduction greatest in transmitter sidelobe versus
  mainlobe region.
• Element-space ACC computation time prohibitive.
  E.g. 2 s to process 372 element array (1x372
  data) for 124 Dwells, 128 Doppler bins t
  124x128x2 sec 8.8 hours!

5
Beamspace Adaptive Channel Compensation

• Motivated by the need to constrain ACC to
  minimize leakage of interference into receive
  directions covered by the transmit mainlobe where
  target detection is performed.
• Beamspace ACC adaptively reconstructs the receive
  beams of the full array in the transmit mainlobe
  region so that strong directional components have
  minimal leakage into adjacent beams, subject to
  the constraint that data distortion at the good
  elements is within some tolerance.
• Beam-space ACC facilitates suppression of
  interferences which are present or have leaked
  into conventional receive beams.
• Because matrices to invert and decompose are of
  dimension equal to the number of beams (e.g. 18)
  vs. the number of elements (e.g. 372), beamspace
  ACC is much faster for large receive arrays.
• For ROTHR data, beamspace ACC is 40 times
  faster than element-space ACC (e.g. what took 4
  hours now takes 6 minutes) and 5 times faster
  than MMSE. Time per range/Doppler cell is 45 ms.
• Choice of tolerance parameter determines the
  trade-off between leakage of mainlobe clutter
  into adjacent beams and signal wavefront
  distortion.

6
03 Aug 2004 Rocket Data Analysis Using BACC

• 03 August 2004 PBIQ data 061555 to 062859
  collected using ROTHR Virginia to illuminate the
  Kennedy space center.
• One of the receive shelters was offline, so 32
  receivers near the middle of the array have gone
  bad.
• The geo-display on the left shows the location of
  the rocket at 061859.

7
Identifying the Faulty Sensors

• Covariance matrix (averaged over all Doppler and
  range bins), indicates there are faulty sensors
  undetected by system.
• BACC software treats all sensors with normalized
  power above or below 3 dB deviation as faulty.

8
Doppler vs. Dwell Displays for Rocket Launch

• Scroll Displays (Doppler vs. Time) at target
  range and azimuth showing the rocket movement
  from 061555 to 061915.
• Conventional (left) and BACC (right)
• Note that BACC shows a clean track of the rocket
  with reduced noise.

9
Scroll Display for Conventional Processing

• Conventional Scroll Display
• Note the sidelobe levels especially at Dwells 6,
  8, 12, 21, 25, 29, 34 are reduced considerably
  using BACC.

10
Scroll Display for Beamspace ACC Processing

• BACC Scroll Display
• Note the sidelobe levels especially at Dwells 6,
  8, 12, 21, 25, 29, 34 are reduced considerably
  using BACC.

11
Conventional vs. BACC Range-Doppler Surface for
Dwell 8

• The Range-Doppler plots for Dwell 8 at 061638
• Conventional (left) BACC (right)
• The location of the rocket is shown with a
  circle.
• The rocket cannot be seen clearly using
  conventional procedure.

12
Power vs. Range and Doppler Plots for Dwell 8

• Range and Doppler cuts at target Doppler, Range
  for Dwell 8 at 061638
• The rocket is shown with the arrows.
• Range cut (left) for conventional (red) and BACC
  (blue) at Doppler -10.78 Hz
• Doppler cut (right) for conventional (red) and
  BACC (blue) at Range bin 41
• Note the 10-15 dB reduction in the sidelobes with
  BACC.

13
Conventional vs. BACC Range-Doppler Surface for
Dwell 21

• The Range-Doppler plots for Dwell 21 at 061756
• Conventional (left) BACC (right)
• The location of the rocket is shown with a
  circle.
• The rocket cannot be seen clearly using
  conventional procedure.

14
Power vs. Range and Doppler Plots for Dwell 21

• Range and Doppler cuts at target Doppler, Range
  for Dwell 21 at 061756
• Range cut (left) for conventional (red) and BACC
  (blue) at Doppler -8.95 Hz
• Doppler cut (right) for conventional (red) and
  BACC (blue) at Range bin 42
• Note the reduced sidelobes and sharper peaks with
  BACC.

15
Conventional vs. BACC Range-Doppler Surface for
Dwell 29

• The Range-Doppler plots for Dwell 29 at 061845
• Conventional (left) BACC (right)
• The rocket cannot be seen at all using
  conventional procedure.
• The sidelobes are reduced more than 10 dB and as
  a result the rocket can be seen with BACC.

16
Power vs. Range and Doppler Plots for Dwell 29

• Range and Doppler cuts at target Doppler, Range
  for Dwell 29 at 061845
• Range cut (left) for conventional (red) and BACC
  (blue) at Doppler -10.58 Hz
• Doppler cut (right) for conventional (red) and
  BACC (blue) at Range bin 42
• Note that the target can be seen clearly with
  BACC.

17
Conventional vs. BACC Range-Doppler Surface for
Dwell 34

• The Range-Doppler plots for Dwell 34 at 061915
• Conventional (left) BACC (right)
• The location of the rocket is shown with a
  circle.
• The rocket can be seen clearly using BACC.

18
Power vs. Range and Doppler Plots for Dwell 34

• Range and Doppler cuts at target Doppler, Range
  for Dwell 34 at 061915
• The rocket is shown with the arrows.
• Range cut (left) for conventional (red) and BACC
  (blue) at Doppler -10.78 Hz
• Doppler cut (right) for conventional (red) and
  BACC (blue) at Range bin 41
• Note that the sidelobes are decreased
  significantly with BACC.

19
Conventional vs. BACC Doppler-Azimuth Surface for
Dwell 29

• Doppler-Azimuth plots for the transmitter
  mainlobe region for Dwell 29
• Conventional (left) and BACC (right)
• Rocket is shown with a circle.
• The sidelobe leakage from the directional
  interference into the other beams is reduced
  using ACC, as a result the target can be seen
  clearly.

20
Conventional vs. BACC Doppler-Azimuth Surfaces
for Dwell 23

• Doppler-Azimuth plots for the transmitter
  mainlobe region for Dwell 23
• Conventional (left) and BACC (right)
• The leakage of the sidelobes from the
  interference is reduced considerably

21
Summary and Conclusion

• Adaptive channel compensation (ACC) of faulty
  sensors important when directional clutter and/or
  interference present.
• ACC performed on a single snapshot of
  post-Doppler processed data which avoids the
  training snapshot requirements of adaptive
  beamforming solutions.
• ACC proposed as a means of compensating the
  entire array so that strong interferers made to
  look more like uncorrelated plane-waves which can
  then be suppressed by conventional beamformer
  shading.
• Unlike element-space ACC, which suppresses the
  sidelobe leakage from all directions, Beam-space
  ACC is focused on the transmitter mainlobe where
  the detection is performed.
• The analysis of 03 Aug 2004 rocket launch data
  shows that BACC suppresses sidelobes as much as
  10-15 dB in some dwells, and improves detection
  of the target significantly.
• Mathematical formulation of BACC can be applied
  to sparse array beamforming with potential
  application to 2-D receive arrays.