3
Current Data Fusion Endeavors

This chapter first provides illustrative examples of the successful use of data fusion by the Department of Defense (DOD) and private industry that may be analogous to the use of data fusion for transportation security. It then summarizes current data integration and data fusion projects initiated in this area by the Transportation Security Laboratory (TSL) of the Science and Technology Directorate of the Department of Homeland Security (DHS S&T). By examining the successes and failures of the DOD and others and building on the current research, the Transportation Security



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 33
Fusion of Security System Data to Improve Airport Security 3 Current Data Fusion Endeavors This chapter first provides illustrative examples of the successful use of data fusion by the Department of Defense (DOD) and private industry that may be analogous to the use of data fusion for transportation security. It then summarizes current data integration and data fusion projects initiated in this area by the Transportation Security Laboratory (TSL) of the Science and Technology Directorate of the Department of Homeland Security (DHS S&T). By examining the successes and failures of the DOD and others and building on the current research, the Transportation Security

OCR for page 33
Fusion of Security System Data to Improve Airport Security Administration (TSA) has a strong foundation for expanding its use of data fusion by employing a more focused, systems engineering approach. DEPARTMENT OF DEFENSE INITIATIVES Within the DOD, data fusion endeavors have concentrated on the development of tracking algorithms based on multiple input sources and on the development of automatic target recognition (ATR). For example, Beugnon and colleagues1 used a two-security-system fusion model and developed adaptive algorithms for the prediction of vehicle tracking in the presence of security system noise. In the field of ATR, the DOD research, development, testing, and evaluation (RDT&E) budget justification explains that “ATR systems improve the capabilities of our armed forces by enabling them to make better use of the information provided by such military sensor systems as radar, laser, infrared, hyperspectral, identification friend or foe, and electronic signal measurement.”2 As the DOD moves toward greater use of data fusion, reports of specific applications have begun to appear in the technical press, although typically these reports lack quantitative data. Over a decade ago, Aviation Week and Space Technology reported the use of a synthesized picture of a battlefield in a laboratory simulation of Joint Surveillance and Target Attack Radar System (JSTARS) and Airborne Warning and Control System (AWACS) operations, using software developed by Mitre Corporation.3 That report indicated that time-integrated displays were “very powerful in terms of showing the operator what is going on.” Even earlier, the U.S. Navy had deployed data fusion systems on the Aegis cruiser that linked the SPY-1 radar system with all radars on ships within a battle group. This data fusion provides commanders with early warning and target tracking capabilities to detect, identify, and engage both surface and air targets effectively. Similarly, Aviation Week and Space Technology described the use of data fusion in network-centric warfare using the Network Centric Collaborative Technology (NCCT) project. The aim of this project was to obtain large improvements in data quality at the cost of only 10 to 25 percent of the cost of a major sensor upgrade. This article quotes information from the NCCT project as follows: One of the system’s features is a “goldmine algorithm” that was developed to correlate what might be two or three equivocal or fleeting contacts if taken individually. But cross-references often can offer a solid target location. With conventional, single-location intelligence systems, up to 90 percent of contacts go unreported because they are considered unreliable. Moreover, the algorithm cuts false alarms almost to zero.4 1 C. Beugnon, T. Singh, J. Llinas, and R.K. Saha. 2000. Adaptive track fusion in a multisensor environment. Pp. 24-31 in Vol. 1, Proceedings of the Third International Conference on Information Fusion, July 10-13. 2 RDT&E Budget Item Justification Sheet. February 2004. Available at http://www.dod.gov/comptroller/defbudget/fy2005/budget_justification/pdfs/rdtande/OSD_BA3/L-30603232D8Z_ATR__R-2(co)_R-2a__Feb_2004.pdf. Accessed January 26, 2007. 3 D. Hughes. 1994. Air Force explores data fusion for Joint STARS. Aviation Week and Space Technology, March 7. 4 D.A. Fulghum. 2002. It takes a network to beat a network. Aviation Week and Space Technology, November 11:28.

OCR for page 33
Fusion of Security System Data to Improve Airport Security The Limited Operational Capability Europe and the successor program, the All Source Analysis System (ASAS), have provided Army commanders with up-to-date fused awareness of the battle space. Brown and colleagues5 describe a formal process that was used to evaluate the effectiveness of the ASAS as a fusion system. While this was an early system, the later deployments of the ASAS in both Gulf wars have demonstrated its operational effectiveness. However, the ASAS can also be used to illustrate the difficulties of a data fusion program. The developers of this system initially struggled with the goal of automatically bringing situational awareness to military commanders from all available intelligence data. This goal proved much too challenging, given the current state of understanding in areas such as estimation theory, machine learning, and statistical decision theory. The ASAS has now been successfully deployed and, as mentioned above, used in two Gulf wars by relaxing the automation requirements to incorporate human-directed and informed processes for data fusion. The U.S. Navy has a Bayesian data reduction algorithm to help with data flow in a network-centric environment. The algorithm works as a “data fusion engine” and can be an “integral part of network centric warfare.”6 Another naval example is the Sensor System Improvement Program for the Navy’s EP-3 Aries II, developed to give a “fused tactical picture of the battlespace.” This had operational testing in September 2004, with more than 16 missions accumulating 128 flight hours, resulting in a “significant improvement in capability over previous versions.”7 During the past several years, the U.S. Army has been conducting a science and technology program currently entitled Advanced Research Solutions—Fused Intelligence with Speed and Trust (ARES-FIST) to develop advanced technologies providing automated support for responding to commanders’ priority intelligence requirements. The ARES-FIST program is illuminating sources of complexity in this problem domain, developing software technologies and prototype applications to advance the state of the art on problem characteristics in data fusion requiring research. It is also developing technologically mature software applications to provide incremental, yet substantial performance gains in areas such as the rapid identification of critical reports and indicators that analysts need to answer priority intelligence requirements and that commanders need to make decisions and take actions (actionable intelligence). The program is also developing software support to intelligently guide the collection of the information most needed to answer critical intelligence requirements. Military systems have also explicitly considered the human decision maker operating on the output from a data fusion system. To move to higher levels of fusion, it must be possible to provide a realistic estimate of current and future status and even to estimate the intent of an entity (e.g., a vehicle) within the battle space. This capability has 5 D.E. Brown, C.L. Pittard, and A.R. Spillane. 1992. ASSET: A simulation test bed for evaluating data association algorithms, Computers and Operations Research, 19(6):479-493. 6 F. Donovan. 2004. Navy develops algorithm technology to sort through net centric data flow. Aerospace Daily and Defense Report, June 24.. 7 GlobalSecurity. 2004. Available at http://www.globalsecurity.org/intell/systems/ep-3-ssip.htm. Accessed January 26, 2007.

OCR for page 33
Fusion of Security System Data to Improve Airport Security become known as aided adversarial decision making.8 How the human can best be interfaced to such fused data systems has been studied empirically.9 ATR provides similar capabilities—specifically: “Improved ATR will enable our forces to handle an ever increasing load of sensory information in the complex situations encountered in the military missions of the future. ATR capabilities are becoming essential to the warfighter, as the services pursue ‘network-centric’ concepts for exploiting sensory imagery and information acquired through large arrays of sensors at all echelons.”10 There has also been considerable investment in the problem of tracking potentially hostile aircraft. The problem has involved fusing data from multiple radars on multiple targets. Details of such work can be found in Bar-Shalom and Li, in Blackman, and in Kameda and colleagues.11 A nonmilitary application is described by Rogova and colleagues in their report on the use of data fusion algorithms for improved traffic flow for crisis management.12 This work assigns network states based on multisource data as a demonstration of decision-level fusion. The network states could range from “normal flow” to “severe congestion” and could be characterized on the basis of the fusion of data from inputs such as the detection of individual vehicles, queues, traffic counts, or traffic types. The Department of Homeland Security can learn from the experiences of the DOD and U.S. allies that have institutionalized an active, layered defense predicated on sophisticated command-and-control intelligence systems. “To respond quickly to rising threats, the United States requires timely and actionable intelligence. Improved human intelligence collection, improved intelligence integration and fusion, improved analysis of terrorist threats and targets, and improved technical collection against potential chemical, biological, radiological, nuclear, and explosive weapons are all critical in this regard.”13 This success of these principles is described by Assistant Secretary of Defense Benjamin Riley as follows: In Afghanistan, U.S. forces found and hit moving targets in minutes by sharing information. In Iraq, national intelligence moved in minutes to a B-1 Bomber that 8 J. Llinas, C. Drury, W. Bialas, and A.C. Chen. 1998. Studies and Analyses of Vulnerabilities in Aided Adversarial Decision Making. AFRL-HE-WPTR-1998-0099. Dayton, Ohio, Air Force Research Laboratory. 9 Ann M. Bisantz, James Llinas, Younho Seong, Richard Finger, and Jiun-Yin Jian. 2000. Empirical Investigations of Trust-Related System Vulnerabilities in Aided, Adversarial Decision Making. Report for the Center for Multi-source Information Fusion. Department of Industrial Engineering, State University of New York at Buffalo, Amherst, N.Y. January. 10 RDT&E Budget Item Justification Sheet. February 2004. Available at http://www.dod.gov/comptroller/defbudget/fy2005/budget_justification/pdfs/rdtande/OSD_BA3/L-30603232D8Z_ATR__R-2(co)_R-2a__Feb_2004.pdf. Accessed January 26, 2007. 11 Y. Bar-Shalom and X-R. Li. 1995. Multitarget-Multisensor Tracking, Principles and Techniques, YBS Publishing; S.A. Blackman. 1986. Multiple Target Tracking with Radar Applications, Artech House, Norwood, Mass.; H. Kameda, S. Tsujimichi, and Y. Kosuge. 2002. Target tracking using range rate measurements under dense environments, Electronics and Communications in Japan, Part 1, Communications 85(3):19-29. 12 G.L. Rogova, P.D. Scott, and C. Lollett. 2005. Higher level fusion for post-disaster casualty mitigation operations. Paper presented at 8th International Conference on Information Fusion, July 25-28. 13 Department of Defense. 2005. Strategy for Homeland Defense and Civil Support, June, p. 11.

OCR for page 33
Fusion of Security System Data to Improve Airport Security hit the meeting place of senior Iraqis. The military proved adept at developing tactical knowledge in information-constrained operations. Now consider this: most state and local agencies that would initially respond to a terrorist attack in the United States do not have compatible abilities to cull knowledge from the resulting flow of on-scene information.14 Overall, the experience of the DOD with data fusion has been one of gradual learning, with successful systems now deployed throughout all of the services. In most cases the initial versions of these systems did not meet expectations or specifications; however, the development, testing, and deployment of these initial attempts informed the later developments of the successful systems. These experiences have motivated the recommendations by this committee for the establishment of a data fusion authority to provide oversight to systems development, for a formal systems engineering approach of the data fusion processes, and for realistic operational testing and feedback from this testing to systems development. Finding: While the DOD has achieved successes in data fusion, information sharing, and networked operations, it has also had numerous unsuccessful programs in these areas. Those involved in transportation security can learn a lot from both the successes and the failures of the DOD. Finding: Improvements can be made in security operations by effectively employing data fusion. These improvements can be accomplished with existing technologies. Experience in the DOD indicates the potential effectiveness of and benefits to security operations from applying data fusion. RESEARCH AND PRIVATE-INDUSTRY INITIATIVES Private industry uses data fusion to increase production, decrease costs, and minimize the need for operator attention during manufacturing activities. Data fusion can be integrated at many different process steps and in a variety of ways, depending on a company’s needs. An example of data fusion needs in private industry can be drawn from the manufacture of computer chips. This manufacturing activity requires more than 200 individual process steps, each of which must be controlled within a well-characterized range to produce a profitable yield of usable chips. For many years, the data from each individual step—for example, regarding film thickness and line width—were monitored individually, even though it was well understood that interaction between the individual steps could compensate for errors in processing. Using straightforward data integration, wafer lots could be tracked as they moved from the beginning of the manufacturing line to final testing. Recently, 14 Benjamin Riley, Assistant Secretary of Defense. 2003. Information Sharing in Homeland Security and Homeland Defense: How the Department of Defense Is Helping. Department of Defense, Washington, D.C. September, p. 1.

OCR for page 33
Fusion of Security System Data to Improve Airport Security manufacturers have moved toward fusing the data from individual steps and using mathematical models to predict the final yield. Instead of a “pass/fail” for a process step, the actual measurement value is recorded, and the target “window” for each subsequent step is adjusted to maximize the final yield. As the ability to fuse data improves with the increased networking of tools in the manufacturing facility, industry is moving away from measuring a physical dimension on a processed wafer; it is moving toward monitoring voltages and impedances on the processing tool during the actual wafer processing, using the same mathematical modeling approach to predict the final yield. In addition to saving measurement and operator time, understanding which process steps have the largest impact on reducing yield allows the manufacturer to focus resources on improving those critical process steps. This increasing amount of data fusion and the move to monitoring more fundamental parameters are possible because semiconductor manufacturers agreed on interface protocols and made the providing of these interfaces a requirement to the sale of manufacturing equipment. No equipment vendor could survive without being able to support all of the common equipment interface protocols. This “system-level” view by the manufacturers has led to the ability to control a complex manufacturing facility centrally, focusing resources on the biggest yield detractors and decreasing the number of operators required to run a semiconductor manufacturing facility. Research conducted at the Center for Embedded Network Sensing of the University of California at Los Angeles has focused on the development of shared databases that allow multiple users and systems to share, manage, and search continuous data streams.15 While there is no formal decision-making process based on these combined data, the data can inform other commercial, industrial, and security efforts. Finding: Private industry has employed data fusion to enhance quality and to improve production and has developed data fusion infrastructure, including interface specifications and data structure, to allow the collection and analysis of information. TRANSPORTATION SECURITY INITIATIVES The TSL has been involved in a number of projects that might inform the design, implementation, and use of data fusion for transportation security. Table 3-1 summarizes these projects and categorizes them by type: infrastructure for data fusion, data integration, or data fusion. Infrastructure projects look at communications, data modeling, database resources, and techniques for data fusion and data integration. Data integration projects have been focused on centrally locating data from multiple sources. The central location could be the terminal for security personnel or a data store. Finally, data fusion projects have considered the combination of data from multiple sources for threat estimates. 15 G. Chen, N. Yau, M.H. Hansen, and D. Estrin. 2007. Sharing Sensor Network Data. Available at http://research.cens.ucla.edu/pls/portal/url/item/2B2EEE5C176148E8E0406180528D260E. Accessed March 8, 2007.

OCR for page 33
Fusion of Security System Data to Improve Airport Security Perimeter Surveillance The Secure Perimeter Awareness Network (SPAN) program combines multiple detection systems designed to provide early warning and alerts for unauthorized access. Essentially, the program takes advantage of the Airport Security Detection Equipment radar to detect unauthorized entry and combines the data from this radar with data from optical and infrared camera security systems. When deployed near facilities close to water, it could also incorporate data from underwater detection systems. The SPAN is to be deployed in Kennedy International Airport in New York City. A related program, the Seattle Airport Project, has as its objective the fusion of ground surveillance radar and intelligent video into a single track for intrusion detection TABLE 3-1 Data Fusion Projects of the Transportation Security Administration Data Fusion Project Description Project Type Command, Control, Communications, Computation, and Intelligence Laboratory Conduct secure network design, development, implementation, and engineering activities to support an evolving architecture for networking of sensors Infrastructure for data fusiona EWR/JAXPORT Vehicle Tracking System, Florida Consists of facility and deployment of a vehicle tracking system in an airport/seaport RF-rich environment to evaluate functional and operational benefits. As of 10/07, this project was suspended due to lack of funding. Infrastructure for data fusion Fusion of Sensors and Systems Evaluate an architecture and design of existing and new commercial-off-the-shelf sensors for perimeter security and stakeholder data distribution. As of 10/07, this project was suspended due to lack of funding but established a test bed being used by Galveston and the Coast Guard. Infrastructure for data fusion Cargo Aircraft Motion Detection/Tracking Demonstrate an integrated motion-detection/camera system on a static aircraft capable of detecting human motion. As of 10/07, the ground portion of this project has been completed. However, the airborne project is ongoing. Data integration Smart Container Adapt Vehicle Access Communicator (VAC) Tracking Unit for use on containers. As of 10/07, this project has been integrated into the EWR/JAXPORT Vehicle Tracking system. Infrastructure for data fusion C3 Checkpoint Podium—PHX Integrate cameras, TRXs, WMDs, ETDs to local C3 Command Center at checkpoint Data integration C3 Checkpoint Podium/RFID Integration—DIA Develop same basic capability as PHX—except selectee carry-on RFID. As of 10/07, this project has been merged with the C3 Checkpoint Podium—PHX Data integration Cargo Information Action Center Consists of virtual network to collect/distribute “Columbia/Snake River stakeholders” data Infrastructure for data fusion SUB-DAX Fusion Fuse sensors in subterranean environments (rail, light rail, vehicular traffic, tunnels) Data fusion Ship Commerce Integrity Fusion of software and models into ship routing/rerouting tool Data fusion NOTE: EWR/JAXPORT, early warning radar/Jacksonville Port Authority, Florida; RF, radio frequency; RFID, Radio Frequency Identification; C3, Command, Control, and Communication; PHX, Phoenix Sky Harbor International Airport, Arizona; TRX, transaction; WMD, weapon of mass destruction; ETD, explosive trace detection; DIA, Denver International Airport, Colorado, SUB-DAX, Subterranean and DAX Technology. a Design provides the basic infrastructure to support future decision and parametric-data fusion.

OCR for page 33
Fusion of Security System Data to Improve Airport Security and identification. An older TSL project, Fusion of Security Systems, employs existing radar technology to provide perimeter defense. The Seattle Airport Project intends to fuse the radar data with data from video systems. The TSL has also developed projects to explore other security approaches for the airport perimeter and aircraft on the ground. The early warning radar/Jacksonville Port Authority Vehicle Tracking System program in Florida will fuse Global Positioning System and radio-frequency identification data to track vehicles for perimeter defense. The Cargo Aircraft Motion Detection/Tracking program is designed to demonstrate an integrated motion-detection/camera system on a static aircraft. The motion detection will direct slewing of camera systems. These projects demonstrate an interest in the use of data fusion to improve perimeter security. Again, they lack a systems approach to their development and common data structures for the extant security systems that would provide the foundation for significant improvements through data fusion. Access-Control Systems An example of a TSL initiative to improve airport access control through data fusion is the Airport Access Control Pilot Program. It is designed to provide access control at intended entry points by integrating data from biometric systems with data from the legacy access-control systems. The goal of this fusion approach is to stop intruders and to provide adequate access control at doorways. The TSA has funded another fusion demonstration project in the access-control area: US Access. This registered-traveler program was created to enable frequent travelers between Dulles International Airport in the Washington, D.C., area and Heathrow Airport in London to go quickly through airport security and immigration control. It will use two fingerprints in an OR logic and fuse them with face recognition in an AND logic. However, the TSL will be allowed to postprocess the data with more advanced fusion logic. In addition, the National Biometrics Security Project will provide data on 10 fingerprints, 9 facial poses, and both irises for 10,000 people. The combination of data obtained through normal business practices plus the additional data should allow for experiments with fusion as a means to enable improved access control; the project has the potential to reduce the burdens of transportation security. Need for a Comprehensive Strategy While the projects described in Table 3-1 provide useful information and results in particular locations, the committee has seen no obvious attempt to develop a comprehensive strategy for the use of data fusion to improve transportation security. Each project is essentially a stand-alone attempt to build localized infrastructure or to share information. There has been no obvious attempt to plan or implement these projects to achieve the most effective use of data fusion at all levels. Finding: The TSL of the DHS S&T has identified the need for applying data fusion and has addressed this need by implementing a number of projects at the system and

OCR for page 33
Fusion of Security System Data to Improve Airport Security checkpoint levels. However, these projects are not the output of a systems engineering analysis (which would involve formal requirements analysis and derivation) of data fusion at all levels: baggage screening, checkpoint, and access control and surveillance. Chapter 4 discusses ways to better implement the projects and other opportunities for data fusion.

OCR for page 33
Fusion of Security System Data to Improve Airport Security This page intentionally left blank.