Decision Support -- Projects

Advanced Problem Analysis, Resolution, and Ranking

Analysis Support to Predictive Battlespace Awareness

Counter-Deception Decision Support

Development of NAS Operational Concepts for the Year 2020 and Beyond

Evaluating Distributed Combat Decision
Algorithms

Improving Capacity of Dual and Triple Converging Configurations

Indications and Warning for Countering Terrorism

Mental Models in Naturalistic Decision Making

Mixed-Initiative Control of Automa-teams (MICA)

Parametric Airspace Design for Future Aviation

Retrospective Source Analysis

Structured Argumentation

Traffic Flow Management: The Next Epoch

Decision Support

Decision Support focuses on cognitive-centered decision support applications and new methods and tools for developing effective systems that support decision-making. Emphasis is placed on decision-making in dynamically changing real-time environments (occurring in a day or less). Research in human decision-making to enable the development of better support systems for the military or other sponsors is covered in this area. Also covered is the demonstration of decision aids that advance the state of the art.

Advanced Problem Analysis, Resolution, and Ranking

Win Heagy, Principal Investigator

Washington

Problem
Advanced problem detection and resolution are part of the integrated sector suite concept for future air traffic management decision support systems (ATM DSS). SWRL (Severe Weather Resolution) is an integral component in the evolution of en route DSS capabilities. Research is needed to validate operational concepts, utility and acceptability, and system benefits.

Objectives
User confidence depends on validation of the “goodness” of the weather products. Research is needed to develop and integrate resolution algorithms (utilizing the weather products) with existing DSS functionality and to determine the operational benefits (efficiency, safety, capacity) of the weather products and resolution algorithms.

Activities
We will hold structured discussions with flight service personnel, aviation weather groups, pilots, and others. We will perform functional analysis of the weather products and continue development of the resolution algorithms. We will also carry out human-in-the-loop experimentation with operational personnel to help validate the resolution logic. Concurrently, we plan to develop methodologies and metrics for measuring performance and benefits.

Impacts
This work will contribute to the state of the art in integrating weather products into DSS automation at the sector level. It will provide for continuity in the evolution of en route DSS capabilities. It has potential for significant improvements in en route efficiency, safety, and capacity. The validation approach promotes cost-efficient utilization of existing assets and supports future technology transfer activities.

Analysis Support to Predictive Battlespace Awareness

Deborah Schuh, Principal Investigator

Bedford and Washington

Problem
Operational modeling and analysis was performed in prior research on the subject of predictive battlespace awareness (PBA). Results indicated a wide range of development opportunities. The most critical need is for analytic tools to support tracking information and predicting future enemy actions. This is a complex type of analysis and not easily supportable by a single “tool” or system.

Objectives
This project seeks to establish a PBA analysis framework, determine the “ground truth” for predictive analysis capabilities through experimentation, and contribute to furthering the state of the art for the various components of the framework. Results from experimentation will be compiled into recommendations on “way ahead” planning and investment strategies for achieving PBA.

Activities
First, we will identify the activities and data sets applicable to predictive analysis as depicted in the PBA operational process and executable models. Next, we will identify capabilities and components that would support those operations together in a technical framework. Finally, we will investigate and experiment with applicable COTS, GOTS, and R&D efforts and evaluate them against this framework.

Impacts
The research will result in a structured approach to developing predictive analysis systems and an investment strategy. This activity is relevant to the USAF’s ongoing PBA definition. In addition, we intend to have a continuing experimental facility and knowledge base to support further research and experimentation opportunities in the areas of integrated intelligence and operations.

Counter-Deception Decision Support

Frank Stech, Co-Principal Investigator

Christopher Elsaesser, Co-Principal Investigator

Washington

Problem
Denial and deception aim to disrupt an adversary's ability to "observe, orient, and decide," and to induce inaccurate impressions about friendly capabilities or intentions, causing the adversary to apply intelligence collection assets inappropriately, or fail to employ capabilities to best advantage. While the need for counter-deception (CD) is recognized, proposed solutions make little or no use of the psychology of deception and decision-making.

Objectives
We will develop a decision framework based on existing research on the psychology of deception, and integrate the framework with belief modeling tools to create a CD decision support system for intelligence analysts. Our hypothesis is that the psychology of decision-making and deception can be combined with existing belief management and planning technology to produce a CD decision support system.

Activities
In the Modeling phase we will construct a psychological framework of deception based on a deception taxonomy and deception cognitive model. In the Development phase we will develop tools for generating deception hypotheses and assessing the evidence of deceptions. The result will be a computational system that helps analysts to recognize potential deception moves, evaluate evidence, identify probable deceptions, and de-bias estimates. The Assessment phase will test the hypothesis through experiments with intelligence analysts.

Impacts
Research in CD will position MITRE to assist in several intelligence community initiatives. The research will position MITRE to develop systems to address several of our sponsor's identified "hard problems." It will also augment the Information Operations Planning Tool ACTD with deception planning aids.

Development of NAS Operational Concepts for the Year 2020 and Beyond

Christopher DeSenti, Principal Investigator

Washington

Problem
Prior MITRE research developed a high-level vision of National Airspace System (NAS) operations for the year 2020 calling for new operating paradigms, facilitated by evolving technology and changes in procedures and roles. This vision needs further definition, particularly in light of vision activities now ongoing in the aviation community. Its relationship to overall NAS performance goals must also be communicated.

Objectives
The project will further refine the NAS future vision and define how underlying paradigm shifts could alter the roles of the operational players and their interactions with the system. To show these concepts and obtain meaningful feedback from stakeholders, rapid prototypes will be developed and demonstrated in MITRE’s laboratories.

Activities
Initial work will focus on concept refinement as a means of collaboratively identifying issues and communicating ideas. Scrutiny of underlying assumptions will provide a clearer understanding of their likely impacts on NAS performance. The resulting refinements will feed the building of interactive prototypes in MITRE’s air traffic management laboratory. These will be used to communicate the concepts and provide a mechanism for moving toward consensus among diverse stakeholders.

Impacts
This work will strengthen MITRE’s contributions to industry-wide efforts to define and understand future NAS concepts of operations. MITRE will be better positioned to determine appropriate areas of focus for research and transition planning for long-term NAS concepts of operation.

Evaluating Distributed Combat Decision
Algorithms

William C. Hobart, Jr., Principal Investigator

Washington

Problem
By collocating adjunct processors with missile defense sensors, features can be extracted and shared over a sensor network. However, if discrimination decision algorithms are executed on each of these adjunct processors, they will receive data in different orders due to varying latencies associated with sensor-to-sensor connections. The order in which data is received impacts the convergence of the decision algorithm.

Objectives
By characterizing the impact of data latencies on the convergence of the distributed decision algorithms used in missile defense, it should be possible to derive techniques to speed this convergence. In addition, an understanding of the behavior could well lead to efficient techniques to prevent divergence or at least mitigate its impact.

Activities
We will develop scenarios to provide a context for evaluating data latencies and decision timelines for integrated, layered ballistic missile defense. Other tasks include characterization of convergence dependencies on data ordering, characterization of baseline performance using a simple mitigation technique, development of more sophisticated techniques to control and prevent divergence of the decision algorithms, and, finally, evaluation of these various techniques.

Impacts
Techniques developed during this effort could greatly increase the effectiveness of distributed decision algorithms to support missile defense discrimination and classification. The project addresses a major concern with the proposed sensor netting architecture, and it could have a significant impact on design decisions as development of command and control, battle management, and communications for integrated, layered ballistic missile defense progresses.

Improving Capacity of Dual and Triple Converging Configurations

Anand D. Mundra, Principal Investigator

Washington

Problem
By combining existing procedures for converging runways with expected final approach speeds and a controller automation tool, significant capacity gains can accrue. The key challenges are determining whether the use of the automation tool is feasible, and how accurate the estimates of final approach speeds can be.

Objectives
This project will develop an understanding of the issues involved in implementing an automation tool, Converging Runway Display Tool (CRDA), to assist controllers in increasing the arrival runway throughput at Chicago O’Hare. It will also assess the accuracy of the estimates of final approach speed 20 minutes before landing.

Activities
We will determine the feasibility of CRDA use at Chicago O’Hare through a series of simulations of specific procedures. Chicago controllers will be involved in these simulations. We will continue to work in collaboration with United Airlines to gather data on the accuracy of landing speed estimates.

Impacts
Several airports, including Chicago O'Hare, Dallas-Ft. Worth and Washington Dulles, could benefit from a procedure that utilizes these capabilities. In the case of Chicago this could amount to an increase of 15% to 30% in the arrival capacity of the airport in instrument weather conditions.

Indications and Warning for Countering Terrorism

Lisa Costa, Principal Investigator

Washington

Problem
The I&W community today relies on judgment and experience to identify threats. Watch centers are always “ON,” with little time to reflect, test alternate hypotheses, or review processes, procedures, and methods. Once threats are identified, prioritization is a judgment call. Vestiges of Cold War I&W remain. Open source information is undervalued and not well understood, preventing analysts from taking advantage of completely new forms of analyses.

Objectives
Our objective is to have significant impact on directly funded projects at DIA, the combatant commands, OSD C3I, and other elements of the intelligence community (IC). We would also like to contribute to efforts in the law enforcement arena and provide threat identification for MITRE initiatives on homeland security.

Activities
We will assess operational I&W processes and methods currently employed in the IC; propose a universal model for terrorist assault, ambush, raid, and precision destruction operations; and validate the model through retrospective and predictive analysis. We will also examine and implement best analytical methods for threat assessment and risk measurement and develop a prototype that supports experimentation.

Impacts
As a result of current deficiencies, the public views the IC as either failing to provide warning (9/11) or overreacting to all types of threats without prioritizing them (e.g., crop dusting planes, banks in the northeast, malls, California bridges, theme parks, etc.). We seek to improve the I&W processes used by the IC and potentially those used by law enforcement.

Mental Models in Naturalistic Decision Making

Kevin J. Burns, Principal Investigator

Bedford and Washington

Problem
Computational models are needed to formalize the strengths and bounds of human thinking so that effective systems can be designed to advise people and automate functions. Computational models are also needed to simulate human behavior so that existing systems can be demonstrated and evaluated with fewer people in the loop.

Objectives
Our method is to measure and model human performance in a synthetic environment that poses prototypical tasks of command and control. These tasks include probabilistic risk assessment, dynamic resource management, and competitive/collaborative engagement.

Activities
Our tool is a micro world game called TRACS that offers the dual advantages of practical relevance and empirical rigor. Relevance is achieved via psychological correspondence between game tasks and real tasks. Rigor is achieved via mathematical comparisons between normative (optimal) performance and cognitive (human) performance.

Impacts
Our results are computational models of how, and how well, people make command and control decisions. These models provide a cognitive-scientific basis for designing systems that improve human-computer performance, and for designing agents that behave like people in large-scale simulations.

Mixed-Initiative Control of Automa-teams (MICA)

Christopher L. Johnson, Principal Investigator

Washington

Problem
Currently, the control of unmanned vehicles requires at least one, and often multiple human operators per vehicle. However, new technologies have inspired new concepts in military operations. Future visions involve large numbers of unmanned vehicles operating semi-autonomously, with small numbers of human operators supervising. Transforming these visions into reality will entail many challenges in autonomous and human supervisory control.

Objectives
The MICA program has two equal areas of focus: control-theoretic techniques for autonomous control, and mixed-initiative techniques for integrating humans into the control process. MICA’s mission is to enable multi-level planning, assessment, and control of distributed, large-scale teams of semi-autonomous forces with shared objectives.

Activities
MITRE is supporting MICA program management by providing expertise on the human-centered design and supervisory control issues involved in controlling autonomous teams. Towards this goal, MITRE is investigating the roles of humans in mixed-initiative control systems, techniques for modeling and evaluating human performance, and design principles which assure that humans remain in optimal control of highly automated environments.

Impacts
Solutions to the MICA problem have the potential to revolutionize future military operations, expanding resources while freeing human personnel from dull, dangerous, and costly tasks. Programs where MICA has direct relevance include Future Combat Systems and Unmanned Combat Aerial Vehicles. Solutions would be applicable across multiple tasks—intelligence, combat, search and rescue—and multiple operational environments—air, land, water, space.

Parametric Airspace Design for Future Aviation

Graham K. Glover, Principal Investigator

Washington

Problem
No tools currently exist to generate a clean-slate design for US airspace. Typically, initial airspace design alternates are crafted as tweaks to the current design. The designs are then modeled and evaluated by MITRE through various tools and expert judgment. Results are subsequently presented to the customer with qualifications regarding the limitations of the evaluation tools.

Objectives
This project will take available knowledge of airspace design guidelines and constraints, and synthesize a set of key components into an automated airspace design tool. The goal is to create a holistic tool that will yield a starting point on an airspace design that is operationally suitable for safe and efficient traffic flow.

Activities
Factors that influence airspace design will be identified and considered for use in this tool. The model to be developed, the Automated Parametric Airspace Designer, will implement a subset of the factors identified in the first portion of the project. Validation and demonstration will consist of a limited model output check and a test using real-world examples.

Impacts
This project will contribute significantly to the state of the art in global airspace design. It will provide tools and techniques to help develop improved airspace designs for the future NAS.

Retrospective Source Analysis

Paul Lehner, Principal Investigator

Washington and Bedford

Problem
Intelligence analysts rely substantially on source "reliability" ratings to determine which classified sources to use as primary information sources. Unfortunately, the overall accuracy and usefulness of sources of different reliability levels have never been measured; nor have the accuracy and usefulness of open sources and classified sources ever been compared.

Objective
The objective is to develop both a method for evaluating source usefulness and an initial assessment of the usefulness of some source types (e.g., local newspapers, international newspapers, etc.). The source types will be evaluated by accuracy (proportion of claims that are accurate), specificity (proportion of claims useful), and timeliness (sources that first report important events).

Activities
The major activity will be a structured retrospective analysis of archived open source (and possibly classified) reports. For specified topics and time periods, historical reports will be rated for accuracy, specificity and timeliness.

Impact
We anticipate that certain open sources will be as accurate and useful as highly rated classified sources. Such a result would provide new and very useful guidance to analysts on how to find and weigh sources in generating an assessment/conclusion.

Structured Argumentation

Brant A. Cheikes, Principal Investigator

Bedford and Washington

Problem
The foundation of a sound judgment is a carefully prepared argument, grounded on evidence and supported by chains of sound reasoning. Backed by DARPA, information technology developers have been creating so-called structured argumentation tools focused on the intelligence mission. While these tools appear to have great potential to improve analytic judgments, there is no hard evidence yet supporting this conclusion.

Objectives
This project aims to rigorously evaluate the costs and benefits of employing structured argumentation tools and methods, and, in particular, to develop and test argumentation methods that demonstrably improve the quality of analytic judgments.

Activities
Our approach will be to develop a program of empirical research—experiments using experienced analysts both as subject-matter experts and as test subjects. Experiments will be based on short, realistic analysis problems. The first year will investigate the effects of explicit argument structuring on analytic products and processes. The second year will investigate the effects on analytic judgments.

Impacts
Our results will be of two kinds: technical papers that help our sponsors decide whether the benefits of structured argumentation methods significantly outweigh the costs, and guidance to the vendor community concerning the design of effective argumentation tools. In the absence of these results, sponsor decisions regarding argumentation technology will continue to be informed by anecdotal opinion and subjective evidence.

Traffic Flow Management: The Next Epoch

Thomas Berry, Principal Investigator

Washington

Problem
The FAA’s Operational Evolution Plan (OEP) serves as the “blueprint” for the evolution of NAS capabilities over the next ten years or so. In the area of TFM, however, there is a lack of vision and industry consensus on procedures, roles and responsibilities, and automation needs in the post-OEP timeframe. This makes it difficult to formulate research and transition plans to move TFM capabilities forward.

Objectives
The objectives of the project are twofold: (1) to develop a vision for TFM procedures, roles and responsibilities, and automation needs for the 2012 timeframe; and (2) to demonstrate this vision through tangible products such as illustrations, animation, mockups, storyboards, and possibly laboratory-based rapid prototypes.

Activities
The project will begin by defining the problem and exploring the scope of TFM within the bounds of our future timeframe. Easier solutions will be demonstrated first, followed by solutions to more difficult problems, particularly those that focus on integration of TFM into the overall ATM environment.

Impacts
The tangible demonstration of future TFM vision concepts will provide a basis for communicating a vision to industry stakeholders, and subsequently refining that vision toward one representing a consensus view of future needs.