Sequential Adaptive MultiModality Target Detection and Classification using

Sequential Adaptive Multi-Modality Target Detection and Classification using Physics-Based Models: Review 03 Statistical modeling, classification, and sensor management DARPA-MURI Review 2003 Alfred Hero Univ. Michigan Ann Arbor

Sequential Adaptive Multi-Modality Target Detection and Classification using Physics-Based Models: Review 03 Target Search Scenario High. Res Spot Scan Low. Res Spot Scan Strip Scan

Sequential Adaptive Multi-Modality Target Detection and Classification using Physics-Based Models: Review 03 Sensor Deployment Architecture

Sequential Adaptive Multi-Modality Target Detection and Classification using Physics-Based Models: Review 03 Research Loci • Image modeling and reconstruction – Markov random field (MRF) polarimetric models (Hory&Blatt) – 3 D Imaging with uncalibrated sensor nets (Rangarajan&Patwari) • Adaptive detection and classification – Pattern matching and modeling (Costa) – Distributed detection and classification (Blatt&Patwari) • Sequential sensor management – Myopic information-driven approaches (Kreucher) – Non-myopic approaches (Kruecher&Blatt) Common theme: adaptive robust non-parametric methods

Sequential Adaptive Multi-Modality Target Detection and Classification using Physics-Based Models: Review 03 Detection: Target or Clutter Alone?

Sequential Adaptive Multi-Modality Target Detection and Classification using Physics-Based Models: Review 03 Detection: Target or Clutter Alone?

Sequential Adaptive Multi-Modality Target Detection and Classification using Physics-Based Models: Review 03 Target Returns Not Additive or Gaussian SNR=0 d. B – – SNR=6 d. B 1 cm x 1 mm plate at 1 m from ground Plate under forest canopy (10 deciduous trees) 2 GHz SAR illumination Aggregate of three look angles (azimuth=35, 45, 55, elev=180)

Sequential Adaptive Multi-Modality Target Detection and Classification using Physics-Based Models: Review 03 Polarimetric Field Modeling and Reconstruction h-pol. incidence • Field Distribution On FDTD Box (2 GHz) v-pol. incidence

Sequential Adaptive Multi-Modality Target Detection and Classification using Physics-Based Models: Review 03 MRF empirical histogram Conditional Markov transition histogram …estimated from training data

Sequential Adaptive Multi-Modality Target Detection and Classification using Physics-Based Models: Review 03 Causal MRF Field Synthesis Causal k. NN predictor: Non-Causal MRF model:

Sequential Adaptive Multi-Modality Target Detection and Classification using Physics-Based Models: Review 03 Example: K-NN MRF Extrapolation

Sequential Adaptive Multi-Modality Target Detection and Classification using Physics-Based Models: Review 03 Non-parametric MRF density estimator • General penalized MRF transition density estimate • y is observed data • parameter b enforces smoothness • function g(f) captures data-fidelity – g(f)=|f|^2: standard L 2 quadratic regularization – g(f)=|f|: L 1 regularization for denoising • w(x): smoothing within and across neighborhoods

Sequential Adaptive Multi-Modality Target Detection and Classification using Physics-Based Models: Review 03 Cartoon illustration of density estimator K-Nearest Neighbors Estimator Penalized MRF transition Density Estimator

Sequential Adaptive Multi-Modality Target Detection and Classification using Physics-Based Models: Review 03 Application: MRF synthesis and denoising g(f)=|f|

Sequential Adaptive Multi-Modality Target Detection and Classification using Physics-Based Models: Review 03 Target Modeling and Classification • Pattern matching in high dimensions – Standard techniques (histogram, density estimation) fail due to curse of dimensionality – Entropic graphs recover inter-distribution distance directly – Robustification to outliers through graph pruning • Manifold learning and model reduction – Standard techniques (LLE, MDS, LE, HE) rely on local linear fits and provide no means of getting at sample density – Our geodesic entropic graph methods fit the manifold globally – Computational complexity is only n log n

Sequential Adaptive Multi-Modality Target Detection and Classification using Physics-Based Models: Review 03 A Planar Sample and its Euclidean MST

Sequential Adaptive Multi-Modality Target Detection and Classification using Physics-Based Models: Review 03 Convergence of Euclidean MST Beardwood, Halton, Hammersley Theorem:

Sequential Adaptive Multi-Modality Target Detection and Classification using Physics-Based Models: Review 03 Pattern Matching

Sequential Adaptive Multi-Modality Target Detection and Classification using Physics-Based Models: Review 03 MST Estimator of a-Jensen Affinity Two well separated Classes Two overlapping Classes

Sequential Adaptive Multi-Modality Target Detection and Classification using Physics-Based Models: Review 03 MST Estimator of Friedman-Rafsky Affinity Two well separated Classes Two overlapping Classes

Sequential Adaptive Multi-Modality Target Detection and Classification using Physics-Based Models: Review 03 Target model reduction • 128 x 128 images of three land vehicles over 360 deg azimuth at 0 deg elevation Courtesy of Center for Imaging Science, JHU • The 3(360)=1080 images evolve on a lower dimensional imbedded manifold in R^(16384)

Sequential Adaptive Multi-Modality Target Detection and Classification using Physics-Based Models: Review 03 Target-Image Manifold

Sequential Adaptive Multi-Modality Target Detection and Classification using Physics-Based Models: Review 03 2 D manifold Embedding Sampling distribution Sampling A statistical sample

Sequential Adaptive Multi-Modality Target Detection and Classification using Physics-Based Models: Review 03 Geodesic Entropic Graph Manifold Learning and Pattern Matching Algorithm • Construct geodesic edge matrix (ISOMAP, C-ISOMAP) • Build entropic graph over geodesic edge matrix – MST: consistent estimator of manifold dimension and process alpha-entropy – MST-Jensen: consistent estimator of Jensen difference between labeled vectors • Use bootstrap resampling and LS fitting to extract rate of convergence (intrinsic dimension) and convergence factor (entropy)

Sequential Adaptive Multi-Modality Target Detection and Classification using Physics-Based Models: Review 03 Illustration for 3 land Vehicles

Sequential Adaptive Multi-Modality Target Detection and Classification using Physics-Based Models: Review 03 loglog. Linear fit to asymptote LS-Soln: d=13 H=120(bits)_

Sequential Adaptive Multi-Modality Target Detection and Classification using Physics-Based Models: Review 03 Distributed Multisensor Estimation and Detection • Distributed M-estimation (Blatt) – – Ambiguity function is often multimodal: local and global M Distributed measurements make local M more difficult We develop method to discriminate between local/global M Use unsupervised clustering and Fisher information matching • Distributed change detection (Patwari) – – Bandwidth and computation constraints Multilayer vs flat store-detect-forward architecture We study perfromance loss due to bandwidth constraints How much information should be sent to what layers?

Sequential Adaptive Multi-Modality Target Detection and Classification using Physics-Based Models: Review 03 Distributed Estimation and Detection Sensor 1 Sensor 2 Sensor N Processing unit Final estimator Flat Sensor Aggregation Architecture

Sequential Adaptive Multi-Modality Target Detection and Classification using Physics-Based Models: Review 03 Distributed M- Estimation Ambiguity function for Cauchy distributed points on a manifold

Sequential Adaptive Multi-Modality Target Detection and Classification using Physics-Based Models: Review 03 A slice of ambiguity function Global maximum Local maxima

Sequential Adaptive Multi-Modality Target Detection and Classification using Physics-Based Models: Review 03 Key Theoretical Result • The asymptotic distribution of M-estimate is (asymptotically) a Gaussian mixture • Parameters Ref: Blatt&Hero: 2003

Sequential Adaptive Multi-Modality Target Detection and Classification using Physics-Based Models: Review 03 Validation of Key Result – QQ-plots M-estimates are clustered into two groups. Each group is centered according to the analytical mean and normalized according to the analytical variance.

Sequential Adaptive Multi-Modality Target Detection and Classification using Physics-Based Models: Review 03 M-estimator Aggregation Algorithm Estimator 1 Estimator 2 Estimation of Gaussian Mixture Parameters (EM) Estimator N Sample Covariance Analysis Aggregation To Final Estimate

Sequential Adaptive Multi-Modality Target Detection and Classification using Physics-Based Models: Review 03 Illustration Model: • 200 Sensors • 100 snapshots per sensor • Snapshots are 1 D Gaussian 2 -mixture • Known covariance • Unknown means • Sensors generate i. i. d. M-estimates of means and forward to central processor Global maximum Local maximum Ambiguity function.

Sequential Adaptive Multi-Modality Target Detection and Classification using Physics-Based Models: Review 03 Local/Global Maxima Discrimination Algorithm Bad estimates Inverse FIM Bad estimates Good estimates Empirical covariance

Sequential Adaptive Multi-Modality Target Detection and Classification using Physics-Based Models: Review 03 Addition of other Discriminants Value-added due to local acquisition and transmission of likelihood values

Sequential Adaptive Multi-Modality Target Detection and Classification using Physics-Based Models: Review 03 Distributed Estimation and Detection Sensor 1 Sensor 2 Sensor 3 Sensor 4 Sensor 6 Sensor 5 Processing unit Final estimator Hierarchical Sensor Aggregation Architecture

Sequential Adaptive Multi-Modality Target Detection and Classification using Physics-Based Models: Review 03 Detection: Flat vs Hierarchical Architecture • ‘Flat’ [Rago, Willett, et al] 1 2 3 4 5 6 7 • Hierarchical, w/ and w/o Feedback 1 2 3 4 5 6 7 • • Each sensor is limited with identical r At low PF, Hierarchical outperforms Flat Optimal 7 -Sensor . 30 0 = r 0. 30 r=. 10 r = 00. 10 r=. 03 r=0 3 0 r = 0. Legend Flat Hier. w/o Feedback Hier. w/ Feedback Optimal 1 -Sensor

Sequential Adaptive Multi-Modality Target Detection and Classification using Physics-Based Models: Review 03 Sequential Adaptive Sensor Management Single-target state vector: • Sequential: only one sensor deployed at a time • Adaptive: next sensor selection based on present and past measurements • Multi-modality: sensor modes can be switched at each time • Detection/Classification/Tracking: task is to minimize decision error • Centralized decisionmaking: sensor has access to entire set of previous measurements

Sequential Adaptive Multi-Modality Target Detection and Classification using Physics-Based Models: Review 03 Sequential Adaptive Sensor Management • Myopic information-based strategies (Kruecher) – – Multi-target tracking capabilities Fully Bayesian approach Non-linear particle filtering with adaptive partitioning Renyi-alpha divergence criterion • Non-Myopic strategies (Blatt&Kreucher) – MDP value function approximations and rollout methods – Bayesian path averaging – Reinforcement feedback and learning

Sequential Adaptive Multi-Modality Target Detection and Classification using Physics-Based Models: Review 03 Sensor scheduling objective function • Prospective value of deploying sensor s at time t: Sensor agility Prediction Retrospective value of deploying sensor s Available measurements at time t-1

Sequential Adaptive Multi-Modality Target Detection and Classification using Physics-Based Models: Review 03 Information-based Value Function • Incremental information gained from data collected from using sensor s. Can be measured by divergence • Requires posterior distributions of future target state X given future Z and given present Z, resp. , • Main issues for evaluation of E[D(s, t)|Z] – Computation complexity – Robustness to model mismatch – Decisionmaking relevance

Sequential Adaptive Multi-Modality Target Detection and Classification using Physics-Based Models: Review 03 Value Function : Alpha Divergence • Properties of Renyi divergence – Simpler and more stably implementable than KL (Kreucher&etal: TSP 03) – Parameter alpha can be adapted to non-Gaussian posteriors – More robust to mis-specified models than KL (Kreucher&etal: TSP 03) – Related directly to decision error probability via Sanov (Hero&etal: SPM 02) – Information theoretic interpretation

Sequential Adaptive Multi-Modality Target Detection and Classification using Physics-Based Models: Review 03 Relevance of alpha-D to Decision Error • Consider testing hypotheses • Sanov’s theorem: optimal decision rule has error • Implication: nearly-optimal decision rule for H 1 is if can generate good estimate of alpha-D

Sequential Adaptive Multi-Modality Target Detection and Classification using Physics-Based Models: Review 03 Multi-Target Bayesian Filtering • Joint multiple target posterior density (JMPD) jointly represents all target states (Kastela) Model Update (Prediction using prior kinematic model) Measurement Update (Bayes Rule) • Update eqns must generally be approximated

Sequential Adaptive Multi-Modality Target Detection and Classification using Physics-Based Models: Review 03 Particle Filter (Metropolis) Approximation • Propose (draw) a set of particles based on some importance (proposal) density q chosen to be as close to the posterior as possible time t-1 • Weight the particles using the principle of importance sampling • Resample particles using above density to avoid degeneracy

Sequential Adaptive Multi-Modality Target Detection and Classification using Physics-Based Models: Review 03 Particle Filtering Illustration Initialize: simulate random samples (particles) from proposal density

Sequential Adaptive Multi-Modality Target Detection and Classification using Physics-Based Models: Review 03 • Model Update: Particle Filtering Illustration Propose new particles from existing particles based on drawing samples from the importance density

Sequential Adaptive Multi-Modality Target Detection and Classification using Physics-Based Models: Review 03 Particle Filtering Illustration Measurement Update: Reweight particles density according to

Sequential Adaptive Multi-Modality Target Detection and Classification using Physics-Based Models: Review 03 Multitarget Tracking: Adaptive Proposals • When targets are well separated in measurement space, each targetpartition of particle evolves independently. • In this case can use independent partition (IP) updates • When targets become “close” target-partitions become dependent • In this case should use coupled partition (CP) updates • Adaptive strategy: use IP unless CP is deemed necessary IP updating CP updating

Sequential Adaptive Multi-Modality Target Detection and Classification using Physics-Based Models: Review 03 Numerical Experiment • Simulation conditions – Linear target motion model: isotropic diffusion – GMTI sensor with dwells over uniform grid – Non-linear return: Rayleigh target and clutter rv – Target detector operates with fixed threshold (Pf=0. 1) – No sensor management

Sequential Adaptive Multi-Modality Target Detection and Classification using Physics-Based Models: Review 03 Tracking Simulated Target Motion w/o SM Sensor makes measurements on a grid The sensor is characterized by a probability of

Sequential Adaptive Multi-Modality Target Detection and Classification using Physics-Based Models: Review 03 Real Target Motion Ten real targets Motion taken from recorded GPS measurements During a battle simulation exercise at NTC.

Sequential Adaptive Multi-Modality Target Detection and Classification using Physics-Based Models: Review 03 Real Target Motion

Sequential Adaptive Multi-Modality Target Detection and Classification using Physics-Based Models: Review 03 Multiple Model for Real Target Motion • Target state vector • Three different models – Target is moving: – Target is stopped: – Target is accelerating:

Sequential Adaptive Multi-Modality Target Detection and Classification using Physics-Based Models: Review 03 Multiple Model for Real Target Motion • Model switching transition matrix

Sequential Adaptive Multi-Modality Target Detection and Classification using Physics-Based Models: Review 03 Tracking Real Multitarget Motion w/o SM Staging area Ten real targets Motion taken from recorded GPS measurements During a battle simulation exercise at NTC.

Sequential Adaptive Multi-Modality Target Detection and Classification using Physics-Based Models: Review 03 Quantitative Results : Adaptive Partitions

Sequential Adaptive Multi-Modality Target Detection and Classification using Physics-Based Models: Review 03 Comparison of Managed and Non-Managed Performance • We illustrate the benefit of info-gain SM with AP implementation of JMPD tracking 10 moving targets. • GMTI radar simulated: Rayleigh target/clutter statistics • Contrast to a periodic (non-managed) scan: same statistics • Coverage of managed and non-managed=50 dwells per second

Sequential Adaptive Multi-Modality Target Detection and Classification using Physics-Based Models: Review 03 Tracker Comparison Managed vs. Non-Managed • Monte Carlo tests (left) show performance with SM using 50 looks similar to periodic scan with 700 looks – SM makes the tracker 12 times as efficient in terms of sensor resources needed. • More extensive runs in similar scenario (right) with 3 targets show performance with SM using 24 looks similar to periodic (non-managed) performance with 312 looks – SM makes the tracker approximately 13 times as efficient in this scenario. – Performance of managed scenario with 24 looks at SNR = 2 (3 d. B) similar to performance of periodic management at SNR = 9 (9. 5 d. B) – approximately a 6. 5 d. B performance gain. Utility of Sensor Management : Three Simulated Targets

Sequential Adaptive Multi-Modality Target Detection and Classification using Physics-Based Models: Review 03 Choice of Alpha: Matched Models • When filter model matches the actual target kinematics very closely, the performance of the algorithm is insensitive to the choice of a. • Simulation: Three targets moving according to a nearly constant velocity model with diffusive component q. Filter has exact model of target motion with correct q. • Results: Tracker performance nearly identical for all values of a.

Sequential Adaptive Multi-Modality Target Detection and Classification using Physics-Based Models: Review 03 Choice of Alpha: Mismatched Models ~ Maximum at (3450, 650) Maximum at (4250, 2450) Snapshot of information map for ten target GPS simulation

Sequential Adaptive Multi-Modality Target Detection and Classification using Physics-Based Models: Review 03 Choice of Alpha: Mismatched Models • Under target kinematic model mismatch using a = ½ yields better performance. • Simulation: Ten targets with trajectories taken from real, recorded data. The filter kinematics are mismatched to vehicles with nearly constant velocity. • Results: Fewest lost tracks over 50 Monte Carlo trials with a=. 5

Sequential Adaptive Multi-Modality Target Detection and Classification using Physics-Based Models: Review 03 Multimode Radar: Mode and Dwell Point Selection • MTI Mode – – Each detection cells is 100 m x 100 m Measures strips 1 x 25 cells long Pd = 0. 9, Pfa =. 001 Detects targets with velocity > MDV • Particle Filter –Multiple model (stopped and moving) –Adaptive Proposal Method – 100 Particles, 3 Targets • Sensor Management –Expected gain for each modality/pointing angle calculated before each measurement. – 12 Looks/time step each of 250 km 2 (total approximately 10% of surveillance area) • FTI Mode – – Measures cells that are 100 m x 100 m Measures spots 5 x 5 cells on the ground Pd = 0. 5, Pfa = 1 e-12 Detects stopped targets

Sequential Adaptive Multi-Modality Target Detection and Classification using Physics-Based Models: Review 03 Myopic vs Non-Myopic Strategies • • Myopic SM computes only one-step ahead Non-myopic SM looks ahead multiple steps Even two step look-ahead can be of value Simple illustration: – Non-myopic information gain criterion – Two targets in two cells – At even time instants only one cell is visible

Sequential Adaptive Multi-Modality Target Detection and Classification using Physics-Based Models: Review 03 Non-Myopic Search Tree 5 0. p= 2. 2 = D A 1. 1 >= D < D S 1 C =. 1 <D> 0 =. 5 0 S 8 C. 1 >=1 <Dc S 2 01 B. 9 >= <D 5 0. . 8 = p 1 = D p= D= 0. 5 0 S 3 S 4 S 9 SA SB <Dd D >=. 0 S 6 S 7 <D> D =. 0 01 p= S 5 C. 1 >=1 <Dc SC SD SE SF <Dc D >=. 001 SG C. 1 =1 Dc> < SH SI SJ <Dd D >=. 0 01 SK

Sequential Adaptive Multi-Modality Target Detection and Classification using Physics-Based Models: Review 03 Non-myopic scheme makes use of this information Myopic scheme uses only this information t rge d a ft T re Le easu M Posterior at t=0, P(X 0|Z 0) One Realization of p(X 2|Z 1) when left target measured Ri gh Me t Ta as rge ure t Prediction at t=1, P(X 1|Z 0) d One Realization of p(X 2|Z 1) when right target measured

Sequential Adaptive Multi-Modality Target Detection and Classification using Physics-Based Models: Review 03 Comparison of Greedy and Non-Myopic (2 step) decision making Myopic: Target lost 22% of the time Non-Myopic: Target lost 11% of the time

Sequential Adaptive Multi-Modality Target Detection and Classification using Physics-Based Models: Review 03 Myopic: Target lost 22% of the time Non-myopic: Target lost 11% of the time

Sequential Adaptive Multi-Modality Target Detection and Classification using Physics-Based Models: Review 03 • Before time 190 (the crossover point): – At even time instants, only one target is visible and the myopic/nonmyopic strategies agree 100% of the time. – At odd time instants, the right method is to measure the right target. The myopic/nonmyopic strategies agree about 85% of the time.

Sequential Adaptive Multi-Modality Target Detection and Classification using Physics-Based Models: Review 03 Foci for 2 nd Year • Non-parametric polarimetric backscatter modeling for multistatic target detection • Target and clutter model reduction and pattern matching • Adaptive non-myopic sensor scheduling and management

Sequential Adaptive Multi-Modality Target Detection and Classification using Physics-Based Models: Review 03 Personnel on A. Hero’s sub-Project(2002 -03) • Krishnakanth Subramanian, 1 st year MS student – Birla Institute of Technology – 50% GSRA • Michael Fitzgibbons, 1 st year MS student – Northeastern Univ. – 50% GSRA • Cyrille Hory, Post-doctoral researcher – University of Grenoble – Area of specialty: data analysis and modeling, SAR, time-frequency

Sequential Adaptive Multi-Modality Target Detection and Classification using Physics-Based Models: Review 03 Personnel on A. Hero’s sub-Project(ctd) • Jose Costa, 3 rd year doctoral student – IST Lisbon – Portugese fellowship, summer GSRA • Chris Kreucher, 3 rd year grad student – UM-Dearborn, Veridian Intl – Veridian support • Neal Patwari, 2 nd year doctoral student – Virginia tech – NSF Graduate Fellowship, summer GSRA • Doron Blatt, 2 nd year doctoral student – Univ. Tel Aviv – Dept. Fellowship, summer GSRA • Raghuram Rangarajan, 2 nd year doctoral student – IIT Madras – Dept. Fellowship, summer GSRA

Sequential Adaptive Multi-Modality Target Detection and Classification using Physics-Based Models: Review 03 Publications(02 -03): Estimation-Classification • J. Costa and A. O. Hero, “Manifold learning with geodesic minimal spanning trees, ” submitted to IEEE T-SP (Special Issue on Machine Learning), July 2003. • A. O. Hero, J. Costa and B. Ma, "Convergence rates of minimal graphs with random vertices, " submitted to IEEE T-IT, March 2003. • J. Costa, A. O. Hero and C. Vignat, "On solutions to multivariate maximum alpha-entropy Problems", in Energy Minimization Methods in Computer Vision and Pattern Recognition (EMM-CVPR), Eds. M. Figueiredo, R. Rangagaran, J. Zerubia, Springer-Verlag, 2003 • D. Blatt and A. Hero, "Asymptotic distribution of log-likelihood maximization based algorithms and applications, " in Energy Minimization Methods in Computer Vision and Pattern Recognition (EMM-CVPR), Eds. M. Figueiredo, R. Rangagaran, J. Zerubia, Springer-Verlag, 2003

Sequential Adaptive Multi-Modality Target Detection and Classification using Physics-Based Models: Review 03 Publications(02 -03): Sensor Management • C. Kreucher, K. Kastella, and A. Hero, “Sensor management using relevance feedback learning, ” submitted to IEEE T-SP, June 2003 • C. Kreucher, K. Kastella, and A. Hero, “Multitarget tracking using particle representation of the joint multi-target density, ” submitted to IEEE T-AES, Aug. 2003. • C. Kreucher, K. Castella, and A. O. Hero, "Multitarget sensor management using alpha divergence measures, ” Proc First IEEE Conference on Information Processing in Sensor Networks , Palo Alto, April 2003. • C. . Kreucher, K. Kastella, and A. Hero, “A Bayesian Method for Integrated Multitarget Tracking and Sensor Management”, 6 th International Conference on Information Fusion, Cairns, Australia, July 2003.

Sequential Adaptive Multi-Modality Target Detection and Classification using Physics-Based Models: Review 03 Publications(02 -03): Sensor Management(ctd) • C. Kreucher, C. , Kastella, K. , and Hero, A. , “Tracking Multiple Targets Using a Particle Filter Representation of the Joint Multitarget Probability Density”, SPIE, San Diego California, August 2003. • C. Kreucher, K. Kastella, and A. Hero, “Information-based sensor management for multitarget tracking”, SPIE, San Diego, California, August 2003. • C. Kreucher, K. Kastella, and A. Hero, “Particle filtering and information prediction for sensor management”, 2003 Defense Applications of Data Fusion Workshop, Adelaide, Australia, July 2003. • C. Kreucher, K. Kastella, and A. Hero, “Information Based Sensor Management for Multitarget Tracking”, Proc. Workshop on Multiple Hypothesis Tracking: A Tribute to Samuel S. Blackman, San Diego, CA, May 30, 2003.

Sequential Adaptive Multi-Modality Target Detection and Classification using Physics-Based Models: Review 03 Publications(02 -03): SP for Sensor Nets • N. Patwari and A. O. Hero, "Hierarchical censoring for distributed detection in wireless sensor networks, ” Proc. Of ICASSP, Hong Kong, April 2003. • N. Patwari, A. O. Hero, M. Perkins, N. S. Correal and R. J. O'Dea, "Relative location estimation in sensor networks, ” IEEE T-SP, vol. 51, No. 9, pp. 2137 -2148, Aug. 2003. • A. O. Hero , “Secure space-time communication, " to appear in IEEE T -IT, Dec. 2003. • M. F. Shih and A. O. Hero, "Unicast-based inference of network link delay distributions using mixed finite mixture models, " IEEE T-SP, vol. 51, No. 9, pp. 2219 -2228, Aug. 2003.

Sequential Adaptive Multi-Modality Target Detection and Classification using Physics-Based Models: Review 03 Synergistic Activities(02 -03) • Veridian, Inc – K. Kastella: collaboration with A. Hero in sensor management, July 2002– J. Ackenhusen: collaboration with A. Hero in mine detection, Oct. 2002– C. Kreucher: doctoral student of A. Hero, Sept. 2002 - • ARL – NAS-SED: A. Hero is a member of yearly review panel, May 2002– B. Sadler: N. Patwari (doctoral student of A. Hero) held internship in distributed sensor information processing, summer 2003 • ERIM Intl. – B. Thelen&N. Subotic: collaborators with A. Hero, Oct. 2002 • Chalmers Univ. , – M. Viberg: A. Hero is Opponent on multimodality landmine detection doctoral thesis, Aug 2003 • EMMCVPR: – “Entropy, spanner graphs, and pattern matching, ” plenary lecture, July 2003

Sequential Adaptive Multi-Modality Target Detection and Classification using Physics-Based Models: Review 03 Cross-Fertilization to Other Sponsors(02 -03) • NSF-ITR – “Modular strategies for internetwork monitoring, ” A. Hero, PI (2003 -2008) • NIH-P 01 – “Automated 3 D registration for enhanced cancer management, ” C. Meyer, PI (2002 -2007) • NIH-R 01 – “Radionucleides: radiation detection and quantification, ” N. Clinthorne, PI (2002 -2005) • Sramek Foundation – “Genetic pathways to diabetic retinopathy, ” A. Swaroop, PI (2002 -2005)