Chapter 2 explained why the Air Force is growing increasingly reliant on IS&T. Consistent with this explanation is the committee’s recommendation that AFOSR significantly increase the Air Force investment in basic research in IS&T, supporting such areas as decision-support systems for all levels of command, real-time control of air and ground systems with or without humans in the loop, human-system interfaces and interactions, and ubiquitous connectivity of commanders, deployed forces, and manned and unmanned systems.
Currently, AFOSR is organized into four directorates: Physics and Electronics, Aerospace and Materials Sciences, Chemistry and Life Sciences, and Mathematics and Space Sciences. The Mathematics and Space Sciences Directorate has a mix of programs that include topics such as dynamics and control, systems software and reliability, artificial intelligence, and information fusion. Nowhere in this breakout of topics does the term IS&T appear. In fact, as noted elsewhere in this report, the IS&T efforts are spread over multiple programs, such as those listed above. As it reviewed the available AFOSR information by topic and by budget, the committee had difficulty in determining the full scope of what could be categorized as IS&T research. While this might not hamper staff within AFOSR and AFRL, it could be an unnecessary hurdle for interfacing with outside organizations.
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 90
Basic Research in Information Science and Technology for Air Force Needs 9 Future Considerations PROGRAM ORGANIZATION Chapter 2 explained why the Air Force is growing increasingly reliant on IS&T. Consistent with this explanation is the committee’s recommendation that AFOSR significantly increase the Air Force investment in basic research in IS&T, supporting such areas as decision-support systems for all levels of command, real-time control of air and ground systems with or without humans in the loop, human-system interfaces and interactions, and ubiquitous connectivity of commanders, deployed forces, and manned and unmanned systems. Currently, AFOSR is organized into four directorates: Physics and Electronics, Aerospace and Materials Sciences, Chemistry and Life Sciences, and Mathematics and Space Sciences. The Mathematics and Space Sciences Directorate has a mix of programs that include topics such as dynamics and control, systems software and reliability, artificial intelligence, and information fusion. Nowhere in this breakout of topics does the term IS&T appear. In fact, as noted elsewhere in this report, the IS&T efforts are spread over multiple programs, such as those listed above. As it reviewed the available AFOSR information by topic and by budget, the committee had difficulty in determining the full scope of what could be categorized as IS&T research. While this might not hamper staff within AFOSR and AFRL, it could be an unnecessary hurdle for interfacing with outside organizations.
OCR for page 90
Basic Research in Information Science and Technology for Air Force Needs Recommendation. Given the Air Force’s increasing reliance on IS&T, its role in enabling revolutionary concepts in military affairs, and the committee’s recommendation to significantly increase the investment in basic research for IS&T, the committee recommends that AFOSR consider identifying IS&T as a major area within AFOSR. In fact, as the investment in that area increases over time, AFOSR should consider organizing a separate directorate for IS&T and related topics. Given the committee’s recommendation regarding human-system interfaces and interactions, it would seem appropriate to also consider alignment of all human-system and human-effects-related research within this separate directorate. RECRUITMENT OF PROGRAM MANAGERS With the recommended increase in funding for IS&T and the corresponding increase in scope of the research investment, AFOSR will be faced with a staffing challenge. This challenge can be viewed as an opportunity to reach out to the IS&T research community to recruit new program managers. The DARPA model for recruiting and rotating program managers has quite successfully made use of the Intergovernmental Personnel Act and the more recent Experimental Personnel Hiring Authority. The former has been used to recruit candidates from universities and nonprofit organizations such as federally funded research and development centers (FFRDCs), while the latter has been used to bring candidates from industry to government for a limited term to participate as program managers in exciting new areas of R&D. These tools have been used at the program manager and higher levels with great success. Aside from attracting the best and brightest from universities, nonprofits, and industry, these tools also rotate program managers out of the agency and back to endeavors of their choice when their term is completed. AFOSR then has an opportunity to both staff the recommended growth in IS&T and maintain the vitality of the area by bringing in new program managers as replacements for those who are rotating out. Recommendation. AFOSR has an opportunity as well as a challenge to staff the recommended increase in funding for research in IS&T. The committee recommends that AFOSR take advantage of mechanisms such as the Intergovernmental Personnel Act to recruit new program managers and, possibly, directorate directors for some if not all of the expanded investment in IS&T. This approach would facilitate continued vitality in this important and growing area of research.
OCR for page 90
Basic Research in Information Science and Technology for Air Force Needs A MECHANISM FOR FOSTERING EXPERIMENTAL RESEARCH IN IS&T The committee recommends that AFOSR encourage and support more experimental science in its IS&T program. While visiting the Air Combat Command in April 2005, the committee was briefed on an interesting approach to collaboration, the Information Operations Innovation Network. This network, while not a research effort, facilitates collaboration and coordination among technologists and operators in the development of experiments and supporting capabilities; it is used for carrying out technology demonstrations of new concepts in response to capability gaps. Because the specifications for team-focused, network-enabled systems are poorly defined both technically and operationally, there is a real need for joint experiments involving R&D staff and Air Force operations personnel working with early prototype systems. The experiments would then help to define the system requirements, and researchers would gain a tangible understanding of Air Force goals and emerging capabilities. The challenges posed by distributed information systems and their underlying science and technology require a new approach, and the committee recommends establishing team-focused experimental environments for R&D, data collection, experimentation, and demonstration of new concepts. Examples of issues that present a serious challenge to both researchers and Air Force operators include these: How to control distributed systems with or without humans in the loop. How to support decision-making by sharing information and intentions. How to fuse and manage heterogeneous information. How to test and evaluate techniques for offensive and defensive information warfare. The committee uses the label “distributed research and experimentation environment” (DREE) to describe a shared computation infrastructure that supports experimentation within a community of researchers. As the name implies, DREEs could be virtual centers (geographically distributed), and they could involve a mixture of physical and simulated assets. In some cases, the scale or complexity of the experimental environment might be too great to emulate in a DREE. In such cases, modeling and simulation might enable experimentation and collaboration among distributed participants through micro world descriptions that are domain faithful but presented in an unclassified environment. Such modeling and simulation capabilities would enable a proactive discovery process through which Air Force IT challenges come into clearer focus. In addi-
OCR for page 90
Basic Research in Information Science and Technology for Air Force Needs tion to their technical goals, DREEs would themselves add to the Air Force’s experience base in team-focused, network-enabled systems and contribute to interactions across various Air Force R&D communities. Such collaborative experimental arrangements are not unprecedented in research. For instance, consider the following from the economics research community:1 Computational laboratories (CLs) are computational frameworks that permit the study of complex system behaviors by means of controlled and replicable experiments. Agent-based computational economics (ACE) is the computational study of economies modeled as dynamic systems of interacting agents. ACE researchers generally conduct their studies in the context of CLs. Research groups frequently cite the need for support to create shared data sets, testbed environments, and other infrastructure that will facilitate access to larger, scalable problems and community sharing. For example, consider the following excerpt from a 2003 report on strategic direction for artificial intelligence research at Cornell University:2 A common theme that emerged in all workshop sessions was the importance of data sets and testbeds. … Data sets and testbeds also have a powerful multiplier effect on research progress. A compelling, freely available data set may motivate hundreds of separate research studies, most by researchers with no connection (funding or otherwise) to the original producers of the data set. Particularly in areas where it is difficult to identify in advance the most promising technologies, the resulting breadth of voluntary effort can be crucial. … Support for the creation of freely available, sharable resources is likely to do more to move forward AI research in areas of interest to the Air Force than any other single action. There are several examples of analogous testbeds already in use in the IS&T research community: Many researchers in robotics are participants in periodic RoboCups.3 These events are exciting, involve a large community of researchers, and provide a challenging shared domain for research in perception, adversarial planning, cooperative agent behavior, and machine learning. 1 Available at http://www.econ.iastate.edu/tesfatsi/acedemos.htm. 2 Available at http://www.cis.cornell.edu/iisi/SRDAI-workshop/srdai-iisi-report-2003.doc. 3 Available at http://www.econ.iastate.edu/tesfatsi/acedemos.htm.
OCR for page 90
Basic Research in Information Science and Technology for Air Force Needs Similar benefits of sharing and leverage were seen in the DARPA spoken language experiments from the early 1990s. A large data set based on military messages was prepared each year, and this was used in a community-wide competition to benchmark progress in spoken language research. DARPA-funded research groups were expected to participate, and leading commercial groups joined in; competition was intense and progress was rapid. The DARPA investment was in creating and providing data sets that were shared with the research community. Sharing in this domain required no expensive testbeds or equipment, only shared data sets and performance metrics. Replication of experiments was encouraged and, in fact, conducted intensely. It created an understanding of how different approaches could achieve community-leading performance according to the metrics. In the domain of network services, PlanetLab4 has evolved as a community-supported, distributed multiuser platform that serves as a testbed for overlay networks. It currently consists of 578 machines hosted by 275 sites in more than 25 countries. Most of the machines are hosted by research institutions, although some are elsewhere (e.g., on Internet2’s Abilene backbone). The key objective of the community-developed and community-shared software is to support distributed virtualization—that is, to allocate a slice of PlanetLab’s networkwide hardware resources to an application. This allows an application to run across all (or some) of the machines around the globe, where at any given time multiple applications may be running in different slices of PlanetLab. The advantage to researchers using PlanetLab is that they are able to experiment with new services under real-world conditions and at large scale. An arrangement offering capabilities would be of great value to the Air Force as it tries to develop a better understanding of the design, operation, and management of large, complex networks and systems and of the incorporation of such systems into Air Force operations. Although PlanetLab is more limited in scope than what the committee has in mind, it is given here as an example that works. These examples have three things in common: Each is directed at an exciting problem area. Each involves a research community that is willing, or encouraged, to share research ideas, experiments in which they investigate those 4 Available at http://www.planet-lab.org.
OCR for page 90
Basic Research in Information Science and Technology for Air Force Needs ideas, and results. Ad hoc teams focused on similar problems spawn new ideas at the boundaries of their own idea sets. Each research community supports an infrastructure that can be shared, which may be as modest as data sets and repositories for reports and shared code. The committee urges AFOSR to work with other parts of the Air Force to establish DREEs as analogous testbeds that will allow researchers and Air Force users to experiment with prototype IS&T concepts and systems. Besides the inherent benefit of adding this experimental component to other IS&T research efforts, such an approach would serve as an intellectual crossroads between the scientific and operational communities in support of the scientific discovery process. The approach also gives researchers something very concrete to work on, perhaps including real Air Force data and proposed system specifications, so they can measure the quality of their solutions. The committee was encouraged by a presentation of the AFRL chief technologist, who described how a C2 wind tunnel could provide a computationally based experimentation framework for exploring new technologies and their operational utility. This is an area of great interest to senior Air Force leadership, which has requested a study in 2006 by the Air Force Scientific Advisory Board on rapid, affordable experimentation. The committee believes that DREEs would be useful for each of the main areas of research covered in this report. A DREE for information management, for instance, would enable the associated community—including universities, AFRL directorates, and perhaps FFRDCs—to create sample data sets and develop associated queries that illustrate how the data are to be integrated. A DREE related to network-centric systems would allow exercises that might generate concrete performance requirements, which are otherwise difficult to identify. While exercises are ongoing, operational Air Force participants could clarify their real, not hypothetical, needs; IS&T applied researchers could investigate engineering issues with the prototype network; basic researchers in IS&T could experiment with fundamental changes (to, for example, communication protocols); and HSI researchers could instrument the experiments and learn from them. The DREE approach to experimental science should not be prohibitive in cost, because the necessary network infrastructure is rapidly falling into place and it may become possible to leverage the investments in testbeds by other AFRL directorates. For example, a research version of the distributed mission training (DMT) environment housed in AFRL’s Human Effectiveness Directorate might support experimental science in areas ranging from control of UAVs to decision-making in real-time envi-