Sjogren - Sensing Surveillance & Navigation - Spring Review 2012

SENSING SURVEILLANCE &
NAVIGATION

07 March 2012

Dr. Jon Sjogren
Program Manager
AFOSR/RSE
Integrity  Service  Excellence Air Force Research Laboratory

9 March 2012 DISTRIBUTION A: Approved for public release; distribution is unlimited.. 1

2012 AFOSR SPRING REVIEW
3001N PORTFOLIO OVERVIEW

NAME: Dr. Jon Sjogren

BRIEF DESCRIPTION OF PORTFOLIO:
Integrated/Multi-transmit Radar for Enhanced Imaging Resolution
Innovative Geo-Location with Tracking, Area Denial and Timekeeping

LIST SUB-AREAS IN PORTFOLIO:
Waveform Design/Diversity Exploitation Adaptive to a Varying Channel
“Fully Adaptive Radar” Sensor Processing including MIMO
Sensing for Object Identification: Analysis and Synthesis of Invariants
Integrated Navigation: GPS-based, Inertial, Terrain-following

DISTRIBUTION A: Approved for public release; distribution is unlimited.. 2

Who was “Sensing Surveillance”?

• 1990: Probability and Statistics; Signal Processing
– NM- Directorate of Mathematical and Information Sciences
– Went from emphasis on “Dependability of Mechanical and Human Systems” to Applied
Functional Analysis, Wavelet theory, Analytic de-convolution, and more Wavelets
– Influences: Louis Auslander, E. Barouch, R.R. Coifman, A.V. Oppenheim
• 1996: Signals Communication and Surveillance
– NM- Directorate of Mathematical and Computer Sciences
– Higher wavelet studies, time-scale, time-frequency transformations, Reduced
Signature Targets, Low Probability of Intercept transmission, Fusion of diverse
sensing modalities (“FLASER”) Gurus: A. Willsky, Ed. Zelnio, S. Mallat
• 2002: Sensing, Surveillance and Navigation
– NM- Directorate of Mathematical and Space (and Geo-) Sciences
– Apply earned mathematical technique and computational/data-handling power:
– Design of wave-forms for transmit diversity, combine sensing and
communication, spectrum maintenance, quantum optics and GPS science
– Big names: A. Nehorai, M. A: Approved for public release; distribution is unlimited..D.H. Hughes
DISTRIBUTION Zoltowski, R. Narayanan, 3

SS&N Goals

• Fully Adaptive Radar and Waveform Design
– Payoff: Spectral Dominance, enhanced Radar resolution, EW Countermeasures
• Advances in Automated/Assisted Target Recognition
– Payoff: Identify airborne, ground-based, occluded, camouflaged and moving `
targets.
• Passive Radar Imaging and “Quantum Entanglement”
– Payoff: Perform “surveillance through clouds” by imaging the light source
(instead of the object), together with photon counting.
• Physically Proven Covert Transmission
– Payoff: Achieve high-rate, covert communication through free-space channel,
based on physical/quantum principles.
• Non GPS-based Navigation and Geo-location
– Payoff: Navigation, location and targeting anywhere, with GPS precision.


Vision
Waveform Diversity
Why?
• Rapidly Dwindling Electromagnetic (EM) Spectrum
• Challenging Environments
• Multi-path Rich Scenarios

Waveform Optimization Simultaneous Multi-Function Frequency Diverse Array
•Designed Waveforms for • Spectrally Efficient Waveform •Adaptive Range
Transmit Adaptivity Design Dependent Beam-patterns
•Interference Suppression • Enabled Multi-mission •Electronic Steering with
•System Constraints Capability Frequency Offsets
• Joint Adaptivity on Transmit
•Inherent Countermeasure
and Receive
Capability
Multitasks
J-B Joseph Fourier
F   SAR
MTI

Comm.
W1(t) W2(t) W (t)

3 ….
Wn(t)

 B0 B1 B0
DISTRIBUTION A: Approved for public release; distribution is unlimited..
Available bandwidth 5

Context and Collaboration

Automatic (Assisted) Target Recognition + 3-D modeling:
ARO/ARL emphasizes underpinnings of landmine detection Adelphi MD
Contacts: Ron Myers, A. Swami, J. Lavery
ONR concentration in statistics of acoustics-based target recognition:
Concentration on mathematical techniques such as “reversed Heat Equation”
Collaborators: B. Kamgar-Parsi, J. Tague, R. Madan, John Tangney
DARPA: Previous MSTARs program known for generation of “real-world data”
followed by:
– Mathematics of Sensing, Exploitation, and Execution (MSEE) led by
Dr. A. Falcone collaboration with R. Bonneau RSL et al
– 3-D Urbanscape/URGENT and parallel “Visi-building”
– FOPEN project ran in 2000‟s, now DARPA ATR is under reassessment
Space-Situation Awareness:
Missile Def Agency Funding for Army Radar at Huntsville, technical tasking to MIT Lincoln
Labs; ICBM detection drives Navy Theater Defense (THAD)
International Efforts: DRDC Ottawa (Noise Radar), Singapore Nat Tech Univ (INS and
Integrated Geo-location/Timekeeping)

Scientific Challenges and Innovations

Waveform Design:
Mutual Information as metric is underutilized especially in multiple I-O context
(Geometric Probability / Entropy), expands upon Kullback-Leibler, Shannon
Inversion of a given Ambiguity Function is critical for interpreting radar returns,
can be addressed by Lie‟s theory of symmetries of systems of Differential
Equations (Conservation Laws)
Selection of the Waveform (Woodward) under time-space channel variation, is Ill-
Posed, can be treated by Functional Regularization (Moscow School)

Distributed Synthetic Aperture Radar:
Propagation of singularities (Wave-Front sets) of linear systems (PDE), studied by L.
Hörmander, M.Sato and Yves Meyer, is critical to the inversion of Fourier Integral
Operators attached to simultaneous Range-Doppler SAR surveillance.

Accurate GPS Interpretation, Distributed Synthetic Aperture
Radar:
Hypothesis tests which identify certain central- and non-central χ2 distributions are based
on F distributions with degrees of freedom related to the number of channels.
This is a Key toward rapidDISTRIBUTION A: Approved for public release; distribution is unlimited..
decision: Are there multi-path effects present? 7

Non GPS-based Navigation: Achieve
“Dependable” Precision Navigation and Timing (PNT)
F. Van Graas, M. u.d. Haag, Ohio Univ

In support of sensing, surveillance, guidance/control in caves,
tunnels, under interference
• (Laser) Scanning for Assured 3-D Navigation of UAV
• 3D Navigation:
– Tight integration of Ladar data with Inertial Measurements,
– Use IMU for data association; Ladar for IMU calibration
• Assurance:
– Measured solution covariance (position and attitude) enables the
implementation of an integrity function,
• UAV Design:
– Hovering sensor platform with a 10-lb payload
(platform functions as a sensor gimbal) IMU and
360 Degree Controller Board
Laser Scanne r


High Latitude Ionosphere Scintillation Studies
Problem statement: Jade Y-T Morton, Miami U Ohio Frequent natural scintillations
Ionosphere scintillation degrades space-based communication, occur at Auroral zone
surveillance, and navigation system performance
Project objective:
1. Establish an automatic scintillation event monitoring and global
navigation satellite signal (GNSS) data collection system at HAARP
HAARP, AK
2. Develop algorithms to estimate scintillating GNSS signal parameters.
3. Develop robust GNSS receiver algorithms to mitigate scintillation effects
180-element HF heating array creates artificial scintillations at
HAARP, AK

Phased heating array beam pattern


Multi-Constellation Multi-Band GNSS Array Data
Collection 0.05

Current experimental setup at HAARP include 4 antennas, 24-hour GLONASS & GPS
0
7/27/2011
Sky View of SV Tracks Mask Angle 0 o
7 commercial/software defined GPS/GLONASS receivers. satellite path over HAARP 24 hours from UTC 2010-10-5 00:00:00 1:00-1:30UTC
1

Successful scintillation events captured and processed on 0.8
-0.05 GLO13
GLONASS satellites and GPS L1, L2, and L5 bands. 0.6 -0.1
0.4
Magnetic
-0.15 zenith
0.2

0
-0.2
-0.2 MZ

-0.4
-0.25

-0.6

-0.8
-0.3 GPS12
GLO23
-1 -0.35
-1 -0.5 0 0.5 1

GPS PRN25 Scintillation showing more severer
-0.4
response on the new L5-0.2
-0.3
life-of-safety signal
-0.1 0 0
1.6
L1
L2CM
1.4 L5I

Normalized Signal Intensity
1.2

1

0.8

0.6
0 20 40 60 80 100 120
Time [s] relative to 20-Jul-2010 03:27:30 UTC

Enablers for Sensing, Surveillance,
Navigation
MOTIVATION CONCEPT / PICTURE
•Realize Theme of “Data to
Decisions” Making Optimal Use of
• Operators are overwhelmed by massive
volumes of high dimensional multi-sensor data
Sensors:
• Challenges
-Efficiently process data to extract inherent Keeping Your Head above
information
- Transform “essential “ information into actionable “Torrents of Data”
decisions

Key Topical Thrust Key Topical Thrust
• Conjugate Gradient method for STAP • Embedded Exponential Family of PDF
• Overcomes curse of dimensionality by novel model • Information integration from disparate
order selection via Krylov subspace sensors for detection and classification
• Computationally efficient implementation of • Breakthrough in Statistical Science: Novel
parametric STAP technique for obtaining sufficient statistics
• Attains “matched filter” performance at • Asymptotically optimal in a weak signal
convergence scenario: minimizes K-L divergence from
reference PDF
• Unifies information theoretic criteria (K-L and CRB) public release; distribution is unlimited..
DISTRIBUTION A: Approved for 11
11

Matched Filter and Conjugate Gradient
Algorithm H. Li, Stevens Inst Techn
M. Rangaswamy RYAP
• Optimum matched filter (MF) for multichannel signal detection:

– Direct matrix inverse is computationally intensive
– Need reduced rank solution to reduce training/complexity requirement
– MF can be obtained by minimizing

which can be solved by iterative solvers including
Bernie Widrow steepest descent or conjugate gradient (CG) methods
– CG uses conjugate-orthogonal directions in searching

and converges in no more than M iterations
Matched-Filter Pioneer
(M is the dimension of w)

Conjugate-Gradient Matched Filter and
Conj - Gradient for Parametric Detection

KASSPER Data
output SINR of CG-MF
vs. # of CG iterations

MCARM data

Simulated data


Detection/Classification
with exponentially embedded family of densities under Kullback-Leibler statistics

• Difficult targets
– Limited training data
– Unknown model
– Dependent measurements
• The EEF combines all the
available information efficiently.
– Sensors are not assumed to
be independent.
– Sensor measurements are
succinctly captured via
sufficient statistics for the
EEF.
– Asymptotic optimality in K-L
divergence


Exponentially Embedded Family
Hypothesis Technique

• The exponentially embedded family (EEF) combines all the
available information in a multi-sensor setting from a statistical
standpoint.
• Create PDF using sufficient statistics from each sensor in a multi-
sensor setting
 p 
pη (x)  exp  i Ti (x)  K ( η)  ln p0 (x) 
 i 1 
Stephen Kay, Univ RI
where Ti (the i-th sensor sufficient statistic
is x)
• The Embedded Family of inputs (EEF) asymptotically minimizes
the “Kullback-Leibler” (K-L) divergence from the true model
• Implementable via convex optimization.
• Applications: model order selection, detection/classification,
intelligent multi-sensor integration.


EEF Effectiveness and Efficiency

Receiver Operating Characteristic Correct Classification versus
under Gen‟l Likelihood Ratio Noise Power

ROC curves for the Generalized
Likelihood Ratio Test versus the Probability of correct classification for both
“clairvoyant” Probability Density methods
Function (PDF)


Indoor Modeling

• DoD Applications of indoor modeling:
– Operational situational awareness of individual soldiers and
common operating picture in complex urban environments
– Enables virtual walk-through and fly-through
– Vizualization of exterior and interior, seamless transition
Fast, Automatic, Photo-Realistic, 3D Modeling
of Building Interiors
• State of the Art for Indoor mapping: Dr. Avideh Zakhor
– Wheeled devices on even, smooth surfacesVideo and Image Processing Lab
University of California, Berkeley

• Existing systems cannot deal with uneven surfaces
such as staircases, and do not generate textured 3D
models

Approach to 3D Indoor Modeling

• Use human operator rather than wheeled devices in order to
map/model uneven surfaces, tight environments 

6 “degrees-of-freedom” recovery
• Challenges:
– Weight/power limitations for human operator with backpack
– Unlike outdoor modeling:
• No GPS inside buildings
• No aerial imagery to help with localization
– Unlike wheeled systems with only 3 degrees of freedom: x, y, & yaw,
– Need to recover six degrees of freedom for a human operator: x,y,z, yaw, pitch, roll


L = laser Data Acquisition
C = camera
L-
L-H1
H = horizontal
HU1 C-
V = Vertical C- C-
L C-
L R
R
L-V1
L-V3
L-V3

L-V2
L-V2

L-S2

OMS
Intersense Applanix

HG9900
HG9900

Laptop

DISTRIBUTION A: Approved for public release; distribution is unlimited..
1919

Markov Random Field Formulation of Texture Alignment
•Cast texture selection and alignment as a labeling problem p1
[Lempistky and Ivanov, 07]:
•Include image transformations to generate more image candidates

N
min( Eq (li )    Es (li , l j ))
i 1 ( i , j )
Quality function Smoothness function

Ed (li )  (Trii  Camli ) 2
m
Es (li  l j )   ( I li ( pk )  I l j ( pk )) 2
k 1

•The minimization is a Markov Random Field (MRF) problem which can
be solved using Graph Cut. [Boykov et al. 01]
Quality and smoothness

Quality only N : number of triangles
m : number of sampling points
Ti : position of the ith triangle
li : image label for the i th triangle
Ili : image with label li
Cli : camera position of image I li
min( Ed  Es )
min( Eq ) p k : the k th sampling points in an edge


Interactive
rendering of a Point cloud
two-storey for staircase

model

21

Technical Challenges
for interactive 3-D indoor modeling

• Need for accurate localization:
– More powerful mathematical approaches to sensor fusion
– Fuse lasers, cameras, IMUs with Kalman & particle filtering
– More robust loop closure detection with scanners and camera
– Mathematical techniques to merge multiple local maps
to generate a global map.
• Surface reconstruction:
– Optimization approaches to water tight surface reconstruction
• Simplify models to reduce size:
– Rendering and interactivity
• Systematic characterization of accuracy
– Volumetric characterization rather than by localizing


Removing Atmospheric Turbulence

Prof. Peyman Milanfar (milanfar@ucsc.edu)
University of California, Santa Cruz

Goal: to restore a single high quality image from the observed
sequence

Alternative: Adaptive Optics

Far More Expensive, Large, and Impractical for Tactical Ground Systems

Reconstruction of Distorted Image

Input video Output frame

Top part of the Water Tower imaged at a (horizontal) distance of 2.4 km
For comparisons see http://users.soe.ucsc.edu/~xzhu/doc/turbulence.htm

Low Probability of: „Detection‟, „Interception‟, and „Exploitation‟
in “Free-Space Optical Communications”
Dr. D.H. Hughes,J. Malowicki,P. Cook, AFRL/RITE

“Alpha-Eta” Coherent State Quantum Data Encryption

Phase modulation illustrations of
the same symbol at two different
times: Laser Light Electric Field Expectation in Coherent State
  
 Es (r, t)   S(r, t)   cos  k t  k r    
 2 

Laser Light in Coherent Quantum States
1   eim i
m 0   eim
m

m
m  Logic Assignments for Phase Modulation of Laser Light
M 1
(0,1)   0
m , 1
m  m even
(0,1)    1
, 0 
m m
m odd


-30 dBm Alpha-Eta Coherent “State Quantum Data Encryption”
(QDE) Stationary Experiment
Snapshot Minimum Objective: Determine feasibility of NuCrypt LLC’s
Received Power phase based Alpha-Eta QDE stationary
-30 dbm transmission through a turbulent atmosphere
Approach: Utilize AOptix “curvature” adaptive
optics terminals to compensate for wave-front
phase distortions over a ten kilometer link
-40 dbm
Result: Successful demonstration of QDE
MARGINAL GOOD
transmission, and decryption inversion over 10 km
15 sec free space link.
Average Received Power: - 22 dbm

Eye Diagrams Illustrating successful decryption of random bit stream
Control (Inside) Actual (Outside)
Example:
Simulated Turbulence Real Turbulence


Target Imaging/Target Recognition
Impact of Synthetic/Analytic Innovations

M. Zoltowski, Purdue U.
• “Designed for Diversity”
• Detect and Exploit different aspects of target
• Design and Employ adaptively Multi-Dimensional Wave-forms in
Multi-Antenna Sensing & Surveillance Systems
• Develop toolkit for matrix treatment of MIMO radar wave-forms
• Multiple-Input/Multiple-Output
• enable performance gains through Philip M. Woodward
“Principles of Radar”
• “transmit” and “receive” diversity
• Adapt the wave-form at transmit source/receiver
Norbert Wiener
“Cybernetics”

“Mr.” Generalized Fourier Transform “Mr.” Designed Transmit Waveform

Ultra-High-Resolution Delay-Doppler Radar
Advances in Radar Waveform Diversity M. Zoltowski, Purdue U.

GOAL:
Ultra-High-Resolution Delay-Doppler Radar in
high noise and clutter environments.

APPROACH:
Employ waveform diversity in conjunction with
4x4 Transmit-Receive Radar System.
Enhanced waveform diversity
4x4 Tx-Rx realization: 4 overlapping beams improves detection performance
formed simultaneously on transmit & receive. as shown in this ROC plot.
Each beam transmitting a different waveform

Employ even & odd parts of Golay Pair forming
4-ary complementary set in conjunction with
novel unitary waveform scheduling and joint Other 4x4 realizations:
matched filtering over multiple PRIs •4 overlapping beams formed
Waveforms transmitted simultaneously simultaneously on transmit &
on 2 different polarizations at 2 spatially RESULTS: receive, with each beam trans-
separated transmit arrays
Enhanced waveform diversity leads to ultra-high mitting a different waveform.
time resolution of closely-spaced targets.

Multiplying by a different complex sinewave for
each 4-PRI further reduces background level
while maintaining unimodularity.

Enhanced waveform diversity with unitary scheduling improves target resolution and detection.

Ultra-High Delay-Doppler Resolution Enabled by
Waveform Diversity with Unitary Scheduling

Two targets spaced by only ¼ chip (rectangular chip pulse shaping): Nine 4-
PRI sets (36 PRIs) with each 4-PRI set multiplied by different DT sinewave

SNR=-10 dB

Sensing Surveillance Navigation
(SS&N) Lab Tasks
S.V. Amphay (RWGI): “Manifold Learning, Information-Theoretic Divergence, and
Dimensionality Reduction across Multiple Sensor Modalities” **

Dr. B. Himed (RYAP) “Radar Waveform Optimization” **

Dr. M. Rangaswamy (RYAP), “The Fully Adaptive Radar Paradigm” **

Dr. L. Perlovsky (RYAT), “Theoretical Foundations of Multi-Platform Systems and Layered
Sensing” **

Dr. J. Malas (RYAS), “Characterization of System Uncertainties within a Sensor Information
Channel” **

Dr. D.H. Hughes (RIGE), “Optical Wireless Communications Research” **

Dr. K. Knox (RDSM), “Improved SSA Imaging by the Application of Compressive Sensing” *

Dr. D. Stevens (RIEG), “Characterization of the Method of Time-Frequency Reassignment” *

Dr. G. Brost (RIGD) “Investigation of Ground-Based Radiometric Characterization of the
Slant-Path Propagation Channel for Millimeter Wave Communications” *
*=New for FY12 **=Renewal for FY12

Sjogren - Sensing Surveillance & Navigation - Spring Review 2012

More Related Content

Similar to Sjogren - Sensing Surveillance & Navigation - Spring Review 2012 (20)

More from The Air Force Office of Scientific Research (20)

Recently uploaded (20)

Sjogren - Sensing Surveillance & Navigation - Spring Review 2012