The Center for Security Technologies

Washington University in Saint LouisRonald S. Indeck, Director

Joseph A. O’SullivanRobert Pless

Outline

• WUSTL• CST• CST Intellectual Thrusts• CST Engineering Testbeds• Potential ARL Collaboration

Washington UniversityCenter for Security Technologies

• Washington University:– USNWR 12, top ten in endowment– 8 Schools, Medical, GWBSSW top 3– SEAS: 7 departments, EE, CSE, SSM

• CST– Interdisciplinary Academic Research Center– Formed early 2002– Foundations/history

• Funding– Individual, group and Center level

• Opportunities for ARL-CST Collaboration

Focus of the Center

Electronic sensing of signals from a range of inputs, their generation, management,

transmission, storage, and processing to provide judgment and control critical to

the security and privacy of people, objects, and intellectual assets

CST Coverage

• Broad spectrum of applications– Not just focused on information assurance

• Synergy between technology and policy– Privacy/public policy as ‘design criteria’

• Includes a variety of ‘attacks’– Security is not only terrorism (e.g., natural disasters)

Faculty Breadth

15 Electrical Engineering10 Computer Science4 System Science and Mathematics1 Medicine5 Arts and Science3 Law2 Social Work

CST External Advisory BoardMr. Earle Harbison (retired President and COO, Monsanto), ChairDr. Massoud Amin (Director of Infrastructure Security, EPRI)Dr. Allen Atkins (VP, Boeing)Dr. Tony Cantu (Director, US Secret Service)Prof. Jerry Cox (Senior Professor, Washington University), ChairCol. Tim Daniel (Director, Missouri Office of Homeland Security)Mr. Will Eatherton (Director, Cisco)Dr. Mark Kryder (CTO, Seagate Technologies; former NSF ERC Director)Mr. Jerry McElhatton (Senior VP, MasterCard International)Dr. Craig Mundie (CTO, Microsoft)Dr. Sharon Nunes (Director, IBM)Mr. Joe Leonelli (Director, Battelle)Ms. Jan Newton (President TX, SBC)Mr. Rob Rambaud (CEO, NetLabs)Dr. Don Ross (Chairman, Ross and Baruzzini: Cernium)Hon. William Webster (retired Director, CIA and FBI)

Technology Expertise and Enablersdevices, design, analysis, computation, electronics, systems ...

• Cheap, even disposable sensors• Inexpensive mass data storage• Fundamental advances in imaging science• Reconfigurable, fast electronics • Affordable, rapid computational processing• Economical, pervasive networking,

including wireless

Application – Fingerprints

• Court ruling– fingerprint recognition not a science

Need new scientific foundationsfor biometrics-based recognition and object recognition

• Consider complete system– optical or ultrasound image– computational algorithms– hardware implementation– management/transmission of data– performance prediction

S & T Intellectual Thrusts

• Sensors: Indeck• Advanced Electronic Systems: Lockwood• Information-theoretic Signal and Image Processing: Snyder• Recognition Theory and Systems: O’Sullivan• Vision for Security: Pless• Distributed and Mobile Systems: Roman• Network and Information Security: Hegde• Detection, Isolation, and Accommodation of Faults: Isidori• Privacy, Public Policy, and Ethics: Kieff

Sensors

– Optical Cameras: custom, high-end, off-the-shelf, frame-rates– Infrared, Multispectral, Hyperspectral Sensors– Biological and Chemical Sensors– X-ray, ultrasound, MRI– Custom sensors such as fingerprint, retinal– Magnetic– Radar– Active, passive

Indeck, Brownstein, Fritts, Fuhrmann, Hayes, O’Sullivan, Pless, Smith, Snyder

Advanced Electronic Systems

– Embedded processing– Reconfigurable electronics– Signal processing and VLSI design– Novel electronics including random

number generation

Lockwood, Chamberlain, Franklin, Fritts, Morley, Richard, Rode, Shrauner

Information-theoretic Signal and Image Processing

– Signal and image processing, deterministic and stochastic

– Data modeling, algorithm development, performance prediction

– Signal and information representation, compression, coding

Snyder, Fuhrmann, Grimm, Hegde, Mukai, O’Sullivan, Pless, Wickerhauser, Xu

Recognition Theory and Systems

– Biometrics-based recognition including face,fingerprint, retinal, DNA, etc.

– Physical signature recognition including magnetic signatures– Fast database searching and recognition system implementation– Data and model compression, signal representation– System implementation considerations: processors, computation,

communication, database design– Data mining, intelligence extraction, situational awareness

O’Sullivan, Byrnes, Chamberlain, Franklin, Indeck, Martin, Grimm, Pless, Wickerhauser

Target Target ClassifierClassifier ââ=T72=T72

Vision for Security

– 3D Scene modeling• known or constrained scenes, unknown scenes• Texture, lighting, BRDF• Deterministic, stochastic• Natural scenery, man-made objects, known objects

– Scene and camera motion, known and unknown– 3D registration– Smart cameras, embedded processing– Optical, infrared, hyperspectral, radar imaging systems

Pless, Fuhrmann, Fritts, Ghosh, Grimm, O’Sullivan, Smart, Smith, Snyder

Privacy, Public Policy, and Ethics

– Societal issues, security-privacy perception and reality– Economic issues, cost-benefit analysis – Legal issues– Technological solutions to privacy issues– Facilitate discourse on technology and its implications– Interact together as a group through common queries– Bilateral interactions with technologists regarding

technological enablers and systems design.

Keiff, Clark, Drake, Drobak, Franklin, Indeck, Martin, May, Rank

Engineering Demonstration Testbeds

• Connect all parts• End-end demonstration

– Biometrics/Physics-Based Recognition Systems Morley– Searching Massive Databases for Critical Information Franklin– Networks of Video Cameras Pless– High Speed Network Security Hegde– Security of the Food and Water Supply Smith

Roles of Privacy and Policy Kieff

Fast, inexpensive searches for changing databases• 100 times faster than conventional searches• Scalable, using conventional drives• Search need not be exact

Wide applicability• Genomics• Intelligence• Images

Searching of Massive Databases

Hard drive

Processor

Memory

I/O Bus

Memory Bus

Reconfigurable hardware

Memory/processing

Networks of Distributed Sensors

• Existing or future sensor networks• Networks of waterway sensors

– Detect pollution, bioterrorism, chemical spills– Communicate problems, source identification– Real-time response to evolving situations

• Networks of Cameras– Video or low-rate, complex or constrained scenes– Local vs. distributed computation, communication,

information extraction, – Dynamically reconfigurable systems

Rapid Analysis – Food Supply

• Airport, pollution, biological

• (Chemical) analysis with inputs from– HSI– optical– mass spectrometry

• Search large data set, perform complex computation

Signal Processing for MultipleSignal Processing for Multiple NoncoherentNoncoherent ArraysArrays

• Exploring ways to extend the benefits of sensor arraysignal processing to multiplenoncoherent arrays, separated in time and/or space

• Astronomical Spatial Spectrum Estimate from Individual Snapshots

• Combined Spatial Spectrum Estimate (10 Snapshots)

Knowledge-Aided Sensor Signal Processing andExpert Reasoning (KASSPER)

Sponsor: DARPA

Incorporate side information such as geographical information systems (GIS) to enhance the performance of GMTI radars.

Proposed system will incorporate elements of:• Space-time adaptive processing• Platform position and orientation estimation from GPS• Sensor array autocalibration• HRR simulation using terrain modeling• Machine learning• Sensor scheduling; adaptive waveforms

Digital Array Scanning Interferometer (DASI)

Snyder, WUSnyder, WU Smith, WUSmith, WU

Center for Imaging Science• Washington University led, M. I. Miller, PI• Consortium of universities: MIT, Texas, Brown,UTEP, Harvard• Over time: added UIUC and Yale, dropped Harvard

• Fundamentals of imaging science• scene and sensor modeling, statistical models• algorithm development, performance analysis• built on face recognition work, fundamental physicsand optimal inference

• Contributions to ATR from radar and FLIR, new performance bounds for inference on groups, developed collaborations with NVESD, AMCOM, ARL• CST: applying analogous methodology to security technologies

Washington UniversityCenter for Security Technologies

• Critical mass in security technologies• Many complementary ongoing projects• Widespread applications• Fundamental scientific issues• Guiding standards• Uniquely integrating privacy issues

Synergistic Collaboration

• Homeland security and force protection• ATR and object recognition• Signal and image processing• Vision, scene modeling and visualization

Scientific and Engineering ResourceScientific and Engineering Resource

Washington UniversityCenter for Security Technologies

Distributed and Mobile Systemssensor, computer, communication, and data networks

– Distributed sensors– Mobile computation– Communication– Distributed data storage

Roman, Cytron, Gill, Xu

Network and Information Security

– Hardware solutions for many control issues– Information assurance– Robust distributed database design– Information hiding, watermarking,

steganography, fingerprinting– Distributed threat detection at network speeds– Network combing

Hegde, Lockwood, Min, O’Sullivan, Xu

Detection, Isolation, and Accommodation of Faults

– Robust control theory– Optimization theory– Dynamic vision

Isidori, Byrnes, Ghosh, Mukai

Selected Center Projects

• Forensics• Sensing (optical, HSI)• Object identification• Voice/face recognition• Airport screening• Satellite imagery, vision• Encryption/data security• Network security• Water/food supply

Target Orientation EstimationTarget Orientation Estimation

Likelihood Model Approach:Likelihood Model Approach:ppRR||ΘΘ,,aa((rr||θθ,,aa) ) -- Conditional Data ModelConditional Data ModelppΘΘ,,aa((θθ,,aa) ) -- Prior on orientation (known or simply uniform)Prior on orientation (known or simply uniform)P(P(aa) ) -- Prior on target class (known or simply uniform)Prior on target class (known or simply uniform)

Target Target ClassifierClassifier ââ=T72=T72

Orientation Orientation EstimatorEstimator θθ=135=135

aa=T72=T72

Given a SAR image Given a SAR image rr, , determine a corresponding determine a corresponding target class target class ââ∈∈AA

Given a SAR image Given a SAR image rr and and a target class a target class ââ∈∈AA, , estimate target orientationestimate target orientation

Target Classification Problem

Motivation• Many reported approaches to ATR from SAR• Performance and database complexity are interrelated• We seek to provide a framework for comparison that:

- Allows direct comparison under identical conditions- Removes dependency on implementation details

2S12S1 T62T62 BTR 60BTR 60 D7D7 ZIL 131ZIL 131 ZSU 23/4ZSU 23/4

Publicly available SAR data from the MSTAR program