Agent-Based Dissemination Hastens Information Stream to Warfighters
Commanders establish policies for management of both content and resources using profiles and transfer agents.
The U.S. Defense Department is turning to a family of software agents that locate, recognize and speed the delivery of critical information to where it is needed most on the battlefield. When minutes and seconds are precious commodities to a warfighter, software transfer agents help manage and expedite the dissemination of badly needed information. These robot-like software tools help leverage the power of the information age.
The value of information cannot be realized until it reaches the person who can use it. Significant advantages can be created through the uninterrupted flow of quality information to support combat operations. Exploiting the power of information, however, depends on what can be collected, how it can be processed, and the extent to which it can be distributed. The key is to give the warfighting commander the ability to manage bandwidth and disseminate information. This approach is where the Defense Department’s information dissemination management (IDM) project begins (SIGNAL, May, page 17).
Both the Defense Information Systems Agency (DISA) and the Defense Advanced Research Projects Agency (DARPA) are involved in IDM efforts. The technology developments of both agencies are being merged to facilitate their application to a wider range of programs. Both agencies have developed tools to handle new information management concepts. Strong ties exist between the technical and operational communities within DISA and DARPA that are involved in IDM activities. These connections help provide a better understanding of emerging warfighter opportunities.
As an example, Capt. Bruce Freund, USN, is DISA’s IDM product team leader. His experience as a surface warfare officer, prior to engineering duty and program management, helps bring an operator’s perspective to the effort. Lt. Col. James R. Barbour, USA, an Apache attack helicopter pilot, is DARPA’s battlefield awareness and data dissemination (BADD) advanced concept technology demonstration (ACTD) program manager. The data dissemination part of BADD involves IDM technologies.
Within DISA, a cross-directorate product management team handles IDM development. Management team members are from directorates such as engineering, program management and logistics. “The product line the team is working on is the culmination of several earlier efforts and the use of commercial products. Bits and pieces of commercial and government technologies and capabilities were applied initially to the Bosnia command and control augmentation (BC2A) effort,” Capt. Freund relates.
“For the first time, this Bosnia effort put high-capacity datalinks into a theater of operations,” the captain maintains. “A key was in managing the products that went into that big pipe, since we have not had large bandwidth capabilities before,” he relates. “The programming of what went into the links for the Balkans was a new experience, requiring new tools to deal with the situation. DARPA’s BADD program has gone further in terms of profiling, of understanding what a unit or an operator requires for the mission.”
Once the software technology in both the DISA and DARPA programs proved its military utility through assessments, Col. Barbour illustrates, “it made sense to merge the efforts and take the best features from both approaches and begin fielding IDM. The engineers from both organizations worked to integrate the technologies,” and the segmented software is working its way through the approval process in the defense information infrastructure common operating environment prior to fielding as IDM Release 3.
DARPA’s investment in BADD research and development over the past five years is approximately $100 million. DISA’s funding over the same period for IDM is more than $50 million. The consolidated IDM effort is transitioning to operations and maintenance funding. But questions remain about which organization will become the executive agency and how it will be implemented. Meanwhile, DARPA is seeking a way to continue research funding and hold together the ACTD engineering team after the program ends September 30. Technical advances from DARPA would be funneled into IDM and other related programs.
Exploiting the power of information requires organizing it into meaningful knowledge contexts, then providing that knowledge reliably and in a timely manner to the decision maker. IDM’s software technologies help combine information into a coherent picture, providing a dramatic increase in its value for the commander, Col. Barbour assures. A demonstration of IDM’s capabilities within a realistic combat scenario was conducted for SIGNAL and clearly showed the value of the technology (see page 23).
The IDM concept centers around a transfer plan to direct the activities of virtual robotic agents. These agents, essentially lines of code, interface between the user’s request for information and the product, Col. Barbour explains. The transfer plan tells a recognition agent what the warfighter needs and the agent constantly seeks to locate it. As the information becomes available on a server, a transfer agent sends it to the warfighter. The warfighter only has to ask for the product once. Every time that product changes or is updated, the new information is obtained by the agent and sent to the user, he claims.
IDM’s goals and metrics call for an order of magnitude improvement in the efficiency of dissemination to be achieved by operationally prioritizing dissemination from unmanaged to managed, Col. Barbour depicts. Faster data dissemination reduces the time needed to deliver information to the user from days and hours to minutes and seconds, and it reduces comprehension time from minutes to seconds, according to the colonel.
“We want to do everything possible to shorten the commander’s decision loop,” Capt. Freund stresses, “to help him become more responsive and reduce latency. To do this requires mechanisms by which priorities and bandwidth allocations can be changed automatically, moving information across the theater without having to pass the word from human to human. This is something we can see in the future.” He adds that software tools will have to deal with the change in operations as the planning phase moves into the execution phase. “Real time to a planner is in hours; real time to the guy being shot at is in milliseconds.”
In addition to the ACTD, early releases of the IDM software are being used in the U.S. Pacific Command based in Hawaii, where applications are being expanded from a pilot service that began in South Korea. Additional capabilities are being added to IDM through the third release of the software. Placing the technology in a theater where operators use it “helps capture how to put the tools to use, what makes sense and establishes where the leverage is,” the captain illustrates.
Information must span different time frames as the user moves from garrison to transit, to arrival in the theater, to engagement with an enemy. “The environment and timing requirements will materially change. The only way we can hope to keep up with this situation is through automation tools,” Capt. Freund offers. “The commander now has very little in the way of tools to help maintain control, other than issuing edicts and expecting people to modify their [bandwidth use] behavior as a result of the order. This might be all right if you are in the one-hour to one-minute category. But it doesn’t help much if you are in that one-millisecond range.”
The commander must establish parameters for who receives which information and when it is delivered, according to Capt. Freund. Understanding and capturing in an automated way the information that a unit or an operator requires to execute a particular mission is essential. The capture of specific data leads to brokering the information to support the warfighter, he explains.
Once it is established that the information exists and that someone needs it to accomplish the mission, there is a delivery management service to perform this function. The IDM approach differs dramatically from current practices in that today many of the routing decisions are made absent from an awareness of the information’s precise content. Calling the robotic transfer agent approach a “new game,” Capt. Freund notes, “IDM is seeking to make decisions based on content awareness.” The software-based transfer agents play a significant role in content awareness.
There are commercial search engines available that help the user become aware of various information products, the captain continues. “You can see them on the Web—search engines that help the user learn about the availability of information that can be requested. What IDM is trying to do is not only to gain awareness of products that a Web browser might be aware of, but locate other products such as video, imagery and serial data that are much broader in scope. Once aware that the products exist, gaining access is not necessarily automatic,” the captain states. “Getting access to these products requires an effort.”
Should the Central Intelligence Agency, as an example, have an analysis of a particular theater or an organization, it does not imply easy access. There is a process by which the need for access to the information is determined, the captain points out. When the existence is established and the information requirement is confirmed, based on the profile, the product is moved to the user. “This is what dissemination management is all about—becoming aware of the products and figuring out who needs which product, matching them up and making the transfer.” Capt. Freund says.
He sees the future of IDM as embedded components of other systems. “My true clients with the software segments are really systems like the global broadcast service, the global combat support system (GCSS) and various service command and control systems. These systems need the capability to move products around in an organized fashion in accordance with the commander’s policy,” Capt. Freund remarks. “I see IDM as segments and components that go into other systems, enabling those systems to handle information in a new way.”
Initially, DISA is fielding IDM technology for the global command and control system (GCCS) mission applications. The software will be installed and operate on GCCS servers and workstations. The integrating contractor for the first two IDM software releases was the Charles Stark Draper Laboratory Incorporated, Cambridge, Massachusetts. The laboratory also was involved with the BC2A effort, bringing that Bosnia experience to DISA and the IDM project. Responsibility for the DISA and DARPA IDM products was assigned to Lt. Col. Marlon Brown, USMC, chief engineer for DISA.
Netscape is another IDM contractor widely recognized for its search engine, Compass, which Capt. Freund calls “the heart of our awareness tool.” Another company is Tympani Development Incorporated, Sunnyvale, California, which builds Atlas, a commercial product providing subscription services. This tool places agents on other systems to look for information and then broker the transfer of updates to fulfill user requirements. The data are automatically sent to the user whenever changes occur. Both Compass and Atlas tools have been in use in the South Korea pilot service since the summer of 1998 “doing awareness, searches, cataloging and subscription functions,” he maintains.
Solers Incorporated, Arlington, Virginia, is another company heavily involved in IDM, developing a wide area assured transport service (WAATS). This technology initially was developed as an upgrade for the BC2A system. IDM uses WAATS as the centerpiece for its management functions. The National Imagery and Mapping Agency’s “NIMA in a box” product (SIGNAL, October 1999, page 17) also uses WAATS technology, which is part of IDM Release 2 software.
The IDM technology is also migrating to the Joint Forces Command’s Joint Battle Center, Suffolk, Virginia, for assessment and movement into concepts of operation. Release 3 of the IDM software will be installed at the Pacific Command this summer as a beta test site. This is the same software package that will begin segment certification for the GCCS. This approach will allow the Pacific Command to expand the use of the pilot services in South Korea.
From a joint warrior interoperability demonstration in 1996 that used direct broadcast satellites, the global broadcast service program and the BC2A effort established the capability to move data using commercial technologies, Col. Barbour observes. DARPA formed the BADD program to look at expanding capabilities over a longer term by developing technology. The battlefield awareness part of that program, or BA, is to provide visualization of streaming video, imagery and tracks, he maintains. The concept quickly emerged to make the capability generic, to move information over any type of network and IDM emerged as part of the data dissemination, or DD, element.
Col. Barbour continues that the battlefield awareness portion will transition to other DISA elements involved in visualization activities. However, other programs will continue to examine various aspects of IDM technology. Contractors supporting DARPA’s BADD program, according to Col. Barbour, are TRW, Fairfax, Virginia; TASC, Reading, Massachusetts; Science Applications International Corporation (SAIC), San Diego; The MITRE Corporation, Bedford, Massachusetts; Autometric Incorporated, Springfield, Virginia; Semcor Incorporated, Mount Laurel, New Jersey; DTRI, San Diego; and Sarnoff Corporation, Princeton, New Jersey.
Col. Barbour is convinced that IDM makes an important difference to the commander. The technology provides “automatic enforcement, ensuring that shared resources can be allocated and reallocated in accordance with mission priorities.” Agent-based dissemination accelerates the flow of required information to the warfighter and provides support through quick, easy and automated retrieval. This approach frees the user to concentrate on other tasks. He also reports that metadata-based tools allow quick organization, understanding and delivery of a higher percentage of relevant information.
Metadata, the language used to describe information, is an important element of IDM, Lawrence J. Levin says. He is a consultant to DARPA and an architect for BADD. Metadata are data about data. As an example, if a long text document contains requested information, recognition agents create a small file with parameters in it. After the file is searched, it is cataloged. From these files, this catalog of metadata files tells transfer agents where to look for data on a given topic.
Software robots, which are very customized and extremely small search engines, are instructed to find specific information and post it in a file for ready access. In whatever background material it scans, the robot recognizes sources and reports them to the user.
From a sample set of sources, DARPA discovered that using current data management procedures—which entail entering metadata attributes manually—only about 11 percent of the metadata were being recorded. The transfer agents rely on the presence of metadata, but people cannot be relied on to enter it correctly or even just to enter it. The solution is to use recognition agents for automated metadata entry. This approach improves the entry to 90 percent and thus the relevancy value of the delivered information, while reducing dependence on the man in the loop, Levin warrants.
Once the robot captures metadata based on the content, network monitoring will reveal how they are used based on content over a given period of time, Levin suggests. This can determine how a commander’s bandwidth is used based on the profile. There are transfer agents throughout various defense and intelligence community organizations. Whenever the agent finds information that meets a profile registered for the user, the information is moved to a local repository.
The user’s request for information begins with a graphical user interface and a pull-down menu. The software leads the user through a process that is similar to the way information requests have been handled in the past. “This approach avoids having to train users for a new system. Users see only the information being pushed to them. Behind the scenes … the agents locate information from various sources—an unmanned aerial vehicle or a signals intelligence feed from almost any source,” Levin confirms.
As the recognition agents continually search for user information, they hand it off to transfer agents, which move it to specific access locations based on established profiles, Levin points out. A generic software toolkit provides file agents, which in turn use World Wide Web agents, imagery transformation services agents and database agents to locate the specific information of interest based on a profile, which is a request for information, he contends.
The toolkit also provides streaming agents, which in turn use message streams and continuous streams. The message streams use track agents and signals intelligence agents, and the continuous streams use unmanned aerial vehicle video agents and Cable News Network video agents, Levin discloses. Once the agents determine the content and tag the information, it becomes metadata. The agents are constantly busy in the background locating information while the user performs other tasks.
The obvious question is how do the agents know what to do. They are informed by coordinators in an IDM domain manager of the user’s requirements, when the information is needed—whether hourly, daily, weekly or as soon as it becomes available—and its destination, Levin confides. The domain manager receives a user profile that describes the primary focus and how the user expresses information needs, the type of information necessary, how important it is, and how quickly it is required.
The IDM technology provides what the user needs through metadata-based profiling, recognition agents and transfer agents that are content aware, Col. Barbour concludes. The commander’s resource allocation and access policy provides the capability to determine who has priority and access to information. Policy and the ability of a user’s profile to schedule the transfer agent’s activities provide information when it is needed. Agents organized in a family tree configuration move the information to where it is needed. On the battlefield, when time is crucial, metadata-based tools allow rapid organization and delivery of information to the right person at the right time.
Laboratory Tests Tactical Impact Using Information Dissemination
Located in a setting of olive drab and camouflage netting is an information management, processing and coordination center. Designed to replicate the operations center of a joint theater force commander, the facility participates in simulated combat operations. Resting on canvas floors, banks of workstations and monitors help conduct demonstrations to assess the operational impact of information dissemination management technology.
The center is located within the headquarters of Booz•Allen & Hamilton Incorporated, McLean, Virginia, to help develop concepts of operations and information management. The laboratory also assesses emerging government and commercial technology, verification and validation of software and its documentation.
The information management, processing and coordination center, or IMPACC, is part of a Defense Information Systems Agency (DISA) contract with the company for information dissemination management (IDM) software efforts. Booz•Allen & Hamilton originally created software applications for broadcast World Wide Web and multicast functions, which subsequently transitioned to Charles Stark Draper Laboratory Incorporated, Cambridge, Massachusetts, for IDM development.
The center also is used to prepare preliminary IDM catalogs and profiles and to conduct training on a limited basis. But where it shines is in demonstrating the effectiveness of IDM technology in a combat scenario. IDM services provide awareness, access, delivery and support to the warfighter. Using the Korean Peninsula as the location for a tactical exercise, company officials demonstrated how IDM works within the context of an attack on friendly forces by aggressors using armor. The company’s IDM program manager is Tom Pfeifer, and the laboratory is managed by Bill Spencer. Booz•Allen & Hamilton provides IDM management support to DISA and is responsible for development, integration and fielding of IDM services.
Once hostile armor is detected, various intelligence functions are brought to bear. Based on the use of IDM transfer agents to obtain critical information in a timely manner, friendly forces maneuver to blunt the attack. This includes the use of air support to attack the armored column. As the warfighting scenario unfolds, IDM’s Release 2 software functions are demonstrated at each step.
It quickly becomes apparent that information management is no longer only a command, control and communications (J-6) function or a communications issue. The use of IDM with direct broadcast systems crosses various staff functions that involve operations (J-3), intelligence (J-2) and logistics (J-4). The commander will need someone with broad staff experience to exploit IDM’s full capability, perhaps the deputy commander or an experienced senior chief warrant officer.
This new IDM capability is forcing commanders to rethink procedures to find faster, easier and less manpower intensive ways of accomplishing the mission. It is too soon to determine whether IDM technology can reduce the number of personnel on a warfighting commander’s staff. However, IDM already is materially altering the tasks they perform.