COURSE # ROO-404
MULTI-SENSOR DATA FUSION AND TRACKING
... comprehensive presentation of the state-of-the-art and the evolution of the sensor/data/information fusion technology and it’s applications in support of DoD missions...
Accurate, efficient and timely management of information is vital for successful decision-making in military operations, national security and, indeed, in all aspects of life. The process of automatically filtering, aggregating and extracting the desired information from multiple sensors and other information sources, and integrating and interpreting data is a rapidly advancing technology, commonly referred to as sensor, data, or information fusion. The power to exploit all relevant information rapidly and effectively is at the core of the Net-Centric Operations (NCO) paradigm and is essential to exploiting “Big Data”.
This course explores the state-of the art in multi-sensor integration, target tracking, identification and situation/threat assessment; familiarizes the participants with the principal components and methods of sensor/data/information fusion systems; and reviews the selection and fusion techniques appropriate to system and mission.
Applications and benefits:
You will benefit by enhancing your understanding of:
- The ways multiple sensors can be employed most effectively to improve situation awareness in response to mission needs
- The state-of the art in multi-sensor integration, target tracking, identification and situation/threat assessment
- The response to uncertain or poorly modeled environments
- Learning sources and data characteristics in “Big Data”
- Exploiting context in enriching understanding data
- Criteria for selecting and adapting fusion techniques most appropriate to specific system and mission needs
- The synergies and opportunities for technology re-use in military, intelligence and commercial applications
- The ways to prepare for, and respond to sudden changes in requirements and in technology opportunities
- Establishing trust in automated fusion techniques as appropriate
- Testing and evaluation of multi-sensor systems; as well as target tracking, identification and situation/threat assessment technologies
- The limitations of fusion, and the susceptibility to countermeasures and other adverse conditions.
Who should attend:
Many current and newly proposed DoD programs that focus on discerning better intelligence and tactical data, employ a wide array of advanced sensors and elaborate data fusion techniques. This course presents the capabilities and limitations of this technology, the underlying concepts, the technology and selected applications, thus rendering it invaluable for System Architects, Systems Engineers / Software Developers, Scientists, Managers, Sensors Designers, Military and Intelligence Officers, Operations Personnel, and other users of multi-sensor data fusion for target detection, classification, identification, and tracking.
There are no specific prerequisites for this course, although a general background in electrical engineering, college level mathematics /statistics are recommended for a better understanding of the concepts presented.
- Basic Concepts
- Introduction and Overview:
- Fundamental concepts, definitions and issues
- Where fusion is being used for improved accuracy, knowledge extraction and robustness
- Functional and Process Models for Data Fusion:
- The JDL Model and variants
- Data Fusion/Resource Management Dual Node Architecture
- Data Fusion Architectures and Design Considerations:
- Centralized vs. distributed fusion architectures
- Integrating fusion and resource management
- Measurement-, plot- and track-level fusion
- Fusion in net-centric architectures
- Maintaining data integrity: tracklets, etc.
- Dealing with Uncertainty:
- Uncertainty in sensor data and in models
- Bayesian, Likelihood, Evidential (Dempster-Shafer), Fuzzy, methods
- Human/Machine Task Allocation and Interfaces:
- Human/machine biases and synergies, display issues, presenting uncertainty, human-in-the-loop data fusion
- Data Alignment
- Multi-Sensor Registration, Calibration and Alignment:
- Impact of biases
- Relative, absolute and model-based alignment techniques
- Registering imagery and non-imagery data
- Normalizing Confidence:
- Characterizing Sensors/Sources
- Fusion with dissimilar sensors (e.g. radar + electro-optics + SAR + GMTI)
- Human-sourced information and “hard-soft fusion”
- Data Association
- Hypothesis Generation:
- Gating and clustering methods in target tracking and scene/situation understanding
- Hypothesis Evaluation:
- Bayesian and other methods
- Hypothesis Selection and Track Management:
- Track initiation, confirmation & deletion
- Nearest neighbor, probabilistic (PDA/JPDA), N-Scan/MHT methods
- Assignment algorithms and hypothesis structures: optimality and efficiency in report-to-track association
- Multi-Target State Estimation:
- Probabilistic Hypothesis Density (PHD), Random Set methods
- Target Identification
- Automatic and Aided Target Recognition:
- Concepts and issues, Target detection, classification and recognition
- State-of-the-art correlation, statistical, neural and syntactical/model methods
- Invariants and indexing methods.
- Target and Scene Understanding:
- Compositional model-driven methods
- Specific advanced system discussion.
- Target Tracking
- Kalman Filter and Other Linear Filters:
- State estimation – least squares and Kalman filtering
- Maneuvering Target Tracking:
- Why Non-Linear Filters are needed
- Extended Kalman Filter, Interacting Multiple Model, switching and other adaptive approaches, Particle Filters
- Comparative assessment.
- Use of Contextual Data:
- ID/Feature-aided tracking
- Trafficability-constrained tracking
- Operational/intent guided tracking
- Situation and Threat Assessment
- Concepts and Issues:
- Representing relationships, situations, contexts and scenarios
- Context discovery and exploitation: how to determine relevant evidence
- How to deal with complex and dynamic situations
- Techniques for Situation and Threat Assessment:
- Link analysis, graph matching, templating, case-based reasoning
- Bayesian and Generalized Belief Networks
- Compositional model-driven methods
- Threat Assessment Methods:
- Agent Response Model (adversary model)
- Capability/ Opportunity/ Intent assessment
- Dealing with Uncertain, Incomplete and Inconsistent Data:
- Case-based reasoning, compositional model-driven methods
- The use of machine-learning techniques (SVM, RVM, ANN)
- Adaptive Information Exploitation
- Integrating Data Fusion Resource Management:
- Functional duality
- Utility/Probability/Cost methods for adaptive data collection, data mining, sensor and model management
- Test and Evaluation
- Test and Evaluation as a Data Fusion Problem:
- Test alignment, track-to-truth association, performance evaluation
- Testing and Evaluating Multi-Sensor Systems:
- Metrics of system performance and mission effectiveness
- Specific system metrics, test environments, data sets
- Case Studies (to be discussed throughout the course)
- Examination of diverse real multi-sensor systems, their design, capabilities, operational use and issues; e.g.
- Sensor-fused military aircraft
- Information fusion in surveillance and air defense systems
- Information fusion in technical intelligence
- Information fusion in test and evaluation
- Others as appropriate
- Summary and Discussion
- Current state of the art, expectations for future capabilities
- Topics and issues raised by the class attendees
- Recommended Texts, software packages, conferences, definition of terms, etc.
About the Instructor
Alan Steinberg is recognized internationally as one of the leading experts in information exploitation and sensor fusion, with over 35 years of experience as a designer, developer and operational user of major electronic combat and intelligence systems. An independent consultant, he was most recently Principal Research Scientist at the Georgia Tech Research Institute. Previously he held senior engineering positions with Lockheed, Litton, TASC, USU/SDL, ERIM and CUBRC, with operational experience as an EW operator/analyst for US Army Security Agency in overseas deployment. He played a major role in the design and development of the family of radars for the Missile Defense Agency. As a member of the JDL Data Fusion Group, he revised the well-known JDL Data Fusion Model. He received the prestigious Mignona Award for outstanding achievement in data fusion. He has served on blue-ribbon panels for the US Government to evaluate and recommend technology developments and the restructuring of the Intelligence Enterprise. He teaches numerous short courses on EW, Radar Data Fusion and Target Recognition.
Course: ROO-404 Duration: 3 Days FEE: $1,499 CEUs: 2.16
Please direct any additional inquiries regarding our courses to Zygmond Turski, Program Director, by e-mail, FAX: (240) 371-4488 or TELEPHONE: (202) 241-6326.
Call toll free 1-800-683-7267 from anywhere in the Continental U.S. or CANADA.
Last modified February 10, 2017.