| 2004 Technology
Symposium > Sensors and Environment
Sensors and Environment
Sensors and Environment researches technologies employed to detect, monitor,
and characterize the environment (terrain, weather, targets, etc.) to
determine position within that environment (geoposition), and to manage,
exploit and disseminate positional data (Geographic Information Systems).
The use of radar, optical, sonic, and multispectral sensors will be
covered.
Biometrics: Calibration and Systems Engineering
Nicholas Orlans, Principal Investigator
Location(s): Washington and Bedford
Problems
Broad government needs for personal identification have led agencies to look to biometrics for solutions. DoD, USGC, and DHS programs are underway. Test programs have established technology performance metrics; however, fielded solutions seldom match laboratory performance. Shortcomings are rarely attributed to any particular technology; large performance gains result from sound integration methods, systems engineering, human factors, and use of standards. Objectives
This research helps bridge the gap between technology testing and real systems through a comprehensive, holistic understanding of current test methods and experimentation with real systems. Project researchers continue to collaborate with companion research projects pursuing future generations of 2D and 3D face recognition and fostering open systems and accurate performance reporting. Activities
This research strives to complete activities that established feasibility in the use of synthetic images and performance modeling. Studies of light models, pose invariance, and 3D aspects of face recognition are being explored. Bayesian belief models are used to provide a high-level, holistic approach to fusion of multi-biometric systems. We validate our models by grounding them to additional sensor environments and software infrastructure. Impact
The previous year's effort established a biometrics scenario testing laboratory and produced two publicly released papers; multiple sponsor briefings have been delivered and awareness of and interest in biometrics work continue to increase. Further development of laboratory capabilities and internal knowledge sharing also continues.
^TOP
Dynamic Scheduling for Command and Control Constellation (C2C)
David Zasada, Principal Investigator
Location(s): Washington and Bedford
Problems
No techniques currently exist to dynamically manage multiple, dissimilar, tactically responsive platforms and sensors.
Objectives
Our objective is to develop techniques for the real-time, dynamic management of multiple, dissimilar platforms and sensors. A key objective is to automate the sensor management process, making it highly machine-assisted, while maintaining human-in-control oversight. We intend to develop strategies that achieve operational requirements and optimize performance of the Command and Control Constellation (C2C). Activities
We will exercise a complete sensor task-post-process-use web, concentrating on BMC2 functions and the sensor system resource manager. We emphasize dynamic multi-platform system retasking in response to time-critical targets and events. We will first consider synthetic aperture radar imagery and moving target indication products derived from space-based radar (SBR) and the Multisensor Command and Control Aircraft (MC2A), and then extend our research to Global Hawk. Impact
This project is intended to build a sound technical basis of dynamic sensor management and retasking capabilities for C2C simulations and acquisitions. It is also intended to illuminate potential problem areas in the emerging task, post, process, and use cycles. Potentially, opportunities exist to transition this work into C2C, MC2A, SBR, Multi-Platform Radar Technology Improvement Program, and Global Hawk programs.
^TOP
Innovative Space-Based Radar Antenna Technology (ISAT)
William McLaren, Principal Investigator
Location(s): Washington
^TOP
Multi-Sensor and Multi-Platform Sensor Exploitation for Combat ID
Walter Kuklinski, Principal Investigator
Location(s): Washington and Bedford
Problems
Most operational multi-sensor and multi-platform surveillance systems lack an analytically tractable approach to target ID or automated target recognition (ATR). Historically, target ID/ATR systems have been developed through empirical approaches, leaving few means for understanding observed system performance or predicting how system performance could be improved by including data from new sensing modalities. Theoretical approaches to target ID/ATR can provide the ability to analyze and predict performance and allow sensor systems developed for one application to be assessed in other problem domains. Objectives
The primary objective of this project is to develop, implement, and evaluate optimal fusion approaches for target ID. These approaches will be developed within a unified analytic framework that will allow them to be readily employed in multiple problem domains. The domains range from ground combat situations, where both non-cooperative and cooperative targets are present, to space-based and airborne multi-platform sensor systems. Activities
Two case studies will be undertaken. The first will continue work done during FY02 on surface target characterization. Hyperspectral-imagery/synthetic-aperture-radar data registration, feature extraction, and detection algorithms will be implemented within the framework being developed by the government for high performance computing. The second case study will quantify the potential utility of high range resolution features for air target characterization. Impact
The innovative multiple sensor fusion approaches developed here will improve performance on tracking and identification of time-critical targets. ISR activities that currently rely exclusively on human decision making, such as combat ID/ATR/ATV, can be improved using a hybrid approach. In this approach multiple sensor data-level fusion products are used by human decision makers to improve CID performance
^TOP
Netted Sensors
Laurens Tromp, Principal Investigator
Location(s): Washington and Bedford
Problems
Netted sensors (NS) can address problems such as asymmetric threats and homeland security. The NS concept is based on the premise that large numbers of sensors, broadly dispersed in an area, can through the power of the network expedite the development of actionable knowledge. However, NS "blurs" the boundaries among traditional subsystem engineering disciplines. As a result, NS requires interdisciplinary R&D. Objectives
The overall objectives are to research and develop technology to make NS viable, build a testbed, and demonstrate technology and NS concepts and solutions relevant to sponsors operational problems. Activities
We perform research on critical technology required to make NS viable (e.g., power management, ad hoc network formation, distributed processing and data management). The project will also develop a testbed environment for NS R&D technology and integrate technology components into the testbed. We will demonstrate NS solutions and concepts in a set of application-oriented experiments that address sponsor-related operational problems. Impact
This project will position MITRE at the forefront of the NS community. It will further develop corporate expertise and advance the development of heterogeneous NS systems. This work will yield new technology, demonstrate NS capabilities, and produce a laboratory tool suite.
^TOP
Netted Sensors Fence for Homeland Defense
Gregory Crawford, Principal Investigator
Location(s): Washington
Problems
Potential terrorists/adversaries can exploit a wide range of airborne vehicles to effectively deliver weapons (nuclear, chemical and biological) against civilian and military targets. Given the sheer number of potential targets in the United States and the difficulty of detecting and discriminating low observable airborne vehicles in realistic environments, there is currently no effective, reliable solution for dealing with this threat. Objectives
We will establish a technical foundation for pursuing operational system concepts by first developing an in-depth understanding of candidate target signature phenomenology in the modalities of interest (i.e., microwave, infrared, and acoustic). We will then demonstrate a proof-of-concept approach for effective detection and discrimination of airborne threats using a ground deployed, low cost, low power, multimodal sensor fence. Activities
We will analyze, design, and implement concepts and algorithms to demonstrate detection and discrimination of candidate targets using netted data from multiple sensor types. We will identify unique characteristics of the target signature phenomenology, leverage these characteristics to detect small targets, estimate their speed and heading, classify them as "threat" or "no threat," and communicate this information to an interceptor. Impact
This project will provide a low cost, low power (potentially disposable) methodology for performing key 24/7 sentry functions to protect critical civilian and military infrastructure from airborne threats. Armed with the established technical feasibility of this approach, developers can pursue operational systems with numerous civilian and military applications.
^TOP
Netted Sensors Program: First Year Challenge Problem
Marcus Glenn, Principal Investigator
Location(s): Washington and Bedford
Problems
A key issue in border monitoring is the detection and characterization of intentional incursions within geographic regions that are not adequately protected by current stand-off surveillance methods. Objectives
Our goal is to demonstrate the ability to defend a line or boundary region using clusters of netted sensors to detect, identify, and correlate vehicular and pedestrian traffic on the MITRE Washington and Bedford campuses. Critical to success is close coordination with the leads of the algorithm development and simulation and hardware testbed efforts to meet the schedule of the challenge problem demonstration.
Activities
We will develop the concept of operations and ensure that signal processing algorithms are integrated into the hardware testbed to demonstrate trip-wire vehicle/person detection and vehicle classification, sensor cueing, and target correlation and association. Initial tracking efforts will be started. The demonstration will serve as the focus for first-year research in critical technology areas important to netted sensors. It will also ensure that the various technology components are integrated to generate a meaningful demonstration that addresses sponsors' operational problems. Impact
Successful completion of this demonstration will yield several technical contributions to MITRE's netted sensor program, including a low duty cycle resource management scheme, spatially distributed signal processing algorithms, in-network processing and fusion, and multi-level signal processing. This will lay the groundwork for additional research efforts and will demonstrate the utility of netted sensors for solving the border monitoring challenge problem.
^TOP
Netted Sensors R&D
Garry Jacyna, Principal Investigator
Location(s): Washington and Bedford
Problems
The netted sensors community is focused on a few "big ticket" items such as low-power processors, self-configuring networks, and wireless communication protocols. Few researchers have addressed the important role of signal processing beyond simple applications, because most collaborative signal processing problems are technically difficult. For example, communication bandwidths do not support centralized processing approaches. This argues for more functionality at the sensor (mote) level, with increased distributed collaboration between local sensors. Objectives
Our research supports the spiral development of new collaborative netted sensor algorithms for the combat ID and urban warfare challenge problems. Near-term efforts will focus on developing sensor, channel, kinematic, and waveform models in support of the testbed infrastructure effort. The majority of the first-year activities will address problems germane to combat ID and urban warfare in the following areas: sensors and packaging, signal/information processing, information management, resource management, and communications and networking. Activities
The work focuses on waveform/sensor channel modeling in support of the first-year concept demonstration effort, distributed detection and classification of acoustic and electromagnetic signatures at the mote and supernode levels, distributed multi-target tracking using particle-based filter methods, and netted sensor optimization approaches using Markov decision theory. Data exfiltration methods and information management concepts will be examined later in the year. Impact
The Technology Program addresses a number of difficult technical problems, including energy-efficient system optimization methods for sensor cueing and tasking and resource-constrained distributed multi-target tracking. This work will position MITRE at the forefront of cutting-edge research in netted sensors and will further develop corporate expertise in an important emerging area.
^TOP
Netted Sensors Testbed Infrastructure
William Dowling, Principal Investigator
Location(s): Washington and Bedford
Problems
The netted sensors (NS) concept is based on the premise that large numbers of sensors, broadly dispersed in an area, can through the power of the network expedite the development of actionable knowledge to solve operational problems. We believe that NS provide significant operational advantages over conventional approaches in such areas as robustness, ease of deployment and replenishment, adaptability through augmentation to meet emerging threats, covertness, and affordability. We seek to prove these benefits using MITRE's NS testbed. Objectives
We will develop the facilities and technical capabilities necessary to support research, development, and evaluation of NS technologies. The NS testbed will have three major components: a simulation testbed used to perform simulations of NS components and systems; a hardware testbed that encompasses both laboratory hardware assets and sensor fields deployed on MITRE's Bedford and McLean campuses; and an information infrastructure that provides an integration framework for managing the data, applications, and services needed to conduct the research. Activities
This multiyear project will build and integrate the testbed components necessary to support research, experimentation, and challenge problem demonstrations. Simulation testbed efforts will focus on developing models to support emulation of NS nodes that generate synthetic signals. We will also develop a gateway between our MATLAB and DIS simulation frameworks to enable prototype sensor algorithms to be included in large-scale simulations. In the hardware testbed we will develop and integrate new sensor types, collect field data using deployable sensors, develop algorithms using live data available through the simulation network, and port prototype algorithms to sensor node platforms. Finally, we will create an information portal as an integrating framework for NS research. The portal will give researchers access to source code, project documents, and data repositories, and will include an application server and applications to command and control the sensor fields and visualize the collected data. Impact
This project will establish an integrated framework for conducting NS research and provide the facilities that will enable MITRE to participate actively in the NS community. While the prototypes resulting from this research process should be of immediate interest to our sponsors, the overall methodology will be of long-term utility in helping our sponsors develop an understanding of the NS trade space and conduct trade-off analyses to support systems engineering efforts and acquisition decisions.
^TOP
Networked Embedded Software Technology (NEST)
Alex Meng, Principal Investigator
Location(s): Washington
Problems
NEST research centers on securing larger spaces, presumably in remote areas. This calls for deployment of 1000 to 100,000 sensor nodes in a self-locating and self-managing manner to detect, classify, track, and report intrusions. NEST will produce customizable solutions by focusing on software technology such as middleware instead of ad hoc hardware solutions. Further research includes service composition and synthesis to generate customizable code for sensor network applications. Objectives
NEST seeks to advance the state of the art of technology for large-scale wireless sensor networks by developing robust middleware services. With middleware technology, NEST demonstrates customizable applications for different problem scenarios through a series of field experiments. NEST also develops fine-grained integration of physical processes of sensing and actuation with computational processes such as monitoring, diagnostics, detection, and tracking in a closed-loop system control. Activities
MITRE provides independent and objective evaluation of NEST field experiments, and technology appraisal. Through a series of rapid-fire field experiments, NEST matures the technology over increasing scale and difficulty. Experiments in 2003 included Shooter Location and Coordinate Sensor Grid for Urban Warfare scenarios, and Line in the Sand, Waking Up Big Brother, and Red Force Tagging for Special Force operations. In 2004, NEST will field 10,000 sensor nodes in an Extreme Scaling experiment. Impact
MITRE provides concrete and tangible evaluation to the NEST Program Manager. In conjunction with its Netted Sensor Initiative, MITRE exchanges its research and knowledge with NEST, especially on the signal processing technology. MITRE also helps NEST to seek technology transition opportunities.
^TOP
Next-Generation Analyst Environment for Geospatial Intelligence
Tom Bartee, Principal Investigator
Location(s): Washington and Bedford
Problems
Significant hurdles exist in achieving the intelligence community vision of geospatial intelligence. Analysts will soon be overwhelmed by the volume of imagery available (including imagery from commercial sources). Additionally, the user interfaces to current softcopy workstations may be improved substantially through the application of various human-computer interface (HCI) techniques such as cognitive task analysis, adaptive interfaces, and alternative input devices. Objectives
We will define a next-generation geospatial intelligence workstation environment that operates in conjunction with analysts' natural work processes and allows analysts to interface with the system utilizing heuristic knowledge. The focus is on two critical areas: improving the usability of imagery analysis interfaces and integrating imagery data with multi-source data and analysts' heuristic knowledge through research in intelligent image queuing. Activities
We will develop the concept of intelligent image queuing to prioritize imagery for search based on the use of image processing algorithms, geographic information system functions, and multi-intelligence sources. We will build an instrumented environment to monitor and analyze the activity of imagery analysts. We will use HCI and cognitive modeling techniques to design and prototype a new geospatial intelligence production environment. Impact
This activity will provide knowledge and prototype hardware/software capabilities in support of geospatial intelligence that will introduce the geospatial intelligence community to new ways of performing imagery analysis. Prototype capabilities may be transitioned through MITRE sponsors for follow-on development work. Additionally, the research will provide a more tangible understanding of the cognitive processes behind imagery analysis.
^TOP
Synergistic Signal Processing Methods for Sensor Fields
Garry Jacyna, Principal Investigator
Location(s): Washington and Bedford
Problems
The signal processing community has been focused on defining network protocols and developing adaptive (mobile) network architectures for netted sensor systems. This overlooks an important component of the problem: collaborative signal processing. Objectives
The objective of this project is to develop robust signal processing algorithms for distributed sensor detection, classification, localization, and tracking in direct support of MITRE's Netted Sensors Initiative. Activities
This project is structured around developing robust acoustic and electromagnetic blind source separation algorithms, "tripwire" parametric and nonparametric detection methods, feature-based classifiers, adaptive beamforming under positional uncertainty, and multitarget distributed particle filter-based tracking. Impact
This project is part of MITRE's larger netted sensors research program, which will position MITRE at the technical forefront of work within the netted sensors community. The ideas advanced in this project are applicable across a broad spectrum of MITRE and potential sponsor work programs in border security, combat ID, and urban warfare. We anticipate that several algorithms will be patentable.
^TOP
Three Dimensional (3-D) Sensor Exploitation
Walter Kuklinski, Principal Investigator
Location(s): Washington and Bedford
Problems
Existing military target recognition systems fail to meet many desired operational objectives. Sensor systems that could provide accurate real-time 3-D information have the potential to revolutionize target recognition. Since targets and their local environments are three dimensional, 3-D sensors coupled with 3-D processing and exploitation algorithms can produce significant gains in target recognition. Objectives
This project will develop a systems-level methodology to design, analyze, and implement 3-D sensor target recognition systems. To evaluate this methodology a prototype 3-D multiple modality sensor system will be constructed. This prototype system will process sensor data in conjunction with 3-D target models and terrain information to reliably recognize targets over broad ranges of obscuration and environmental complexity. Activities
Statistical signal processing methods based on scattering phenomenology will be applied to the problem of 3-D target recognition. 2-D SAR image formation methods, at both foliage-penetrating VHF/UHF frequencies and millimeter wavelengths, will be extended to provide a 3-D capability. This project will also develop adaptive multi-sensor tasking procedures to optimally task multiple sensor platforms for enhanced 3-D imaging applications. Impact
Sensor systems that provide accurate real-time 3-D information have the potential to revolutionize the target recognition process. The application of advanced signal processing algorithms and intelligent exploitation of 3-D image data in conjunction with a priori information will lead to systems resistant to camouflage, concealment and deception, and jamming, providing increased situation awareness to the warfighter.
^TOP
Urban Vision System
Nick Donnangelo, Principal Investigator
Location(s): Washington
^TOP
Vegetation Forensics
Sherry Olson, Principal Investigator
Location(s): Washington
Problems
Nefarious activities are often extremely difficult to detect directly with today's sensor technology, due to the clandestine and transient nature of activities as well as active denial and deception techniques employed. Indirect sensing techniques may provide the most benefit in some cases. Objectives
Research shows that environmental pollutants, as well as oil, salt, and metals, affect plants in ways that can be measured both in the laboratory and with remote sensing. The stress to plants can be measured after single events or after long-term exposure. This research will demonstrate the application of indirect sensing of vegetation stress stemming from activities of national security interest. Activities
We will conduct plant biology experiments on healthy and stressed vegetation to characterize the effects of stress agents on vegetation under varying conditions. We will collect laboratory and field signatures of the vegetation being studied and conduct remote sensing experiments using this ground truth data. Laboratory, field, and remote spectral data will be analyzed to determine the detection limits and the ability to distinguish between types of stresses caused by natural, nefarious, and benign activities. Impact
Indirect sensing of indicators, such as vegetation stress, has the potential to have a large impact on difficult problems susceptible to denial and deception. Counter-drug applications and other national security concerns where direct sensing of activities range from difficult to extremely difficult are prime candidates. Transition opportunities for this vegetation stress research will be pursued with national and military intelligence organizations.
^TOP
|
