Page 82
C
Synopsis of the CSTB Report Computing and Communications in the
Extreme
The Computer Science and Telecommunications Board's (CSTB's) report Computing and Communications in the Extreme: Research for Crisis Management and Other Applications1 too some initial steps in exploring the role of information technology in crisis management. Discussions at the workshops convened by the steering committee for the project spanned many aspects of computing and communications technology research, development, deployment, and use, focusing on crisis management as the primary application area. The workshops generated ideas about where high-performance technology may be helpful, where advances in performance at the leading edge would yield benefits in more mainstream systems, and how the interaction of applications in different areas (e.g., the use of telemedicine and digital libraries in crisis response) influences the development and use of advanced computing and communications.
Promising computer science and engineering research topics were identified in discussions between crisis management experts and information technologists at the workshops and were developed further in subsequent deliberations by the authoring steering committee. Research in these areas has "the potential to increase the ability of individuals and organizations to make the most of important applications, to present in
1Computer Science and Telecommunications Board, National Research Council. 1997. Computing and Communications in the Extreme: Research for Crisis Management and Other Applications. National Academy Press, Washington, D.C.
Page 83
tellectually stimulating challenges for researchers, and to promote significant advances in the state of technology," according to the report (p. 7).
Research topics suggested in the report include communications resources such as rapidly deployable, self-configuring wireless networks for coordinating response teams; "judgment support" tools to assist crisis managers in making decisions in the absence of complete, reliable information; simulations of phenomena such as hurricanes and fires that could deliver useful results to crisis managers rapidly; and virtual "anchor desks" that would place network-based resources such as simulations and information systems at the disposal of crisis managers.
The steering committee developed 11 findings based on input from the workshops and additional, related information. The findings have a number of common themes. One is that some of the greatest technical challenges stem from the sheer scale (i.e., numbers of people and devices, diversity of resources, amount of computing power, complexity of interactions) of the requirements that must be met. Another theme is that the technologies must be easy enough to use to complement the users rather than distract them from their missions.
The report observes that the widespread interconnection of computing and information resources has made it feasible, and increasingly common, for resources to be called on in unforeseen ways. "Crisis management, in particular, illustrates the value of being able to integrate highly diverse resources whose usefulness in an unusual situation could not have been anticipated in advance. Unfortunately, technologies developed to meet a specific application requirement often do not function well in unforeseen circumstances because of complex, difficult problems of interoperation, performance, and scaling up," the report notes (p. 6). Consequently, the steering committee's findings suggest R&D and deployment efforts that can lead to both architectural approaches for systems that function on a national scale and general-purpose tools and services that facilitate rapid, ad hoc integration of systems and resources.
In developing the findings, the steering committee identified several characteristics of crisis management that place particular stress on adoption and exploitation of advanced information technologies:
• Magnitude. Crises can overwhelm available resources. For example, communications systems, power plants, hospital systems, and weather centers can all be saturated in a crisis. How can systems be developed that have "surge capacity" or that can respond usefully while in a saturated state?
• Urgency. Rapid response in communication and information services to the special loads of crises is essential. It can lead, for example, to communication architectures that provide priority service for crisis man-
Page 84
agers. There can also be a kind of ''engineering urgency," in which interoperation must be established between systems in a matter of hours to satisfy the information needs of responders.
• Infrequency and uncertainty. Some high-magnitude events, such as earthquakes, occur infrequently and in unpredictable locations. How can agencies manage resources to be able to respond effectively, given the reality of constrained budgets? Can information systems be architected to be usable despite the unpredictability of demand?
• Special information needs. A crisis can cause unusual demands both for information flows to and from the crisis area and for consolidation or fusion of information to meet responder needs. For example, in the Oklahoma City bombing of April 1995, information about the Murrah building and nearby buildings was collected from many sources and then "fused" to form a composite model that helped identify high-probability locations to find missing people.
The CSTB report identifies and elaborates on research needs motivated by crisis management. In addition, the report developed a set of findings regarding the role of research in crisis management and other national-scale applications of high-performance computing and communications technology:
• Crisis management testbeds. Finding 1 emphasizes the value of establishing experimental testbeds for crisis management-related R&D. Such testbeds provide a venue for government, academic, and industrial researchers to work with application users, such as federal, state, and local crisis managers, to test and validate technologies by subjecting them to realistic applications. This was the principal finding relating to the process of collaboration between the two communities.
• Infrastructure baselining. Finding 2 highlights the importance of investigating the design and operation of existing national-scale infrastructures to identify which features enable these systems to be scalable, to accommodate diverse components, and to evolve over time.
• Usability and collaboration. Findings 3 and 4 suggest ways to improve support for the human-computer interface and for human activities mediated by computing. The findings call for research to gain a better understanding of users' needs, capabilities, and limitations, as well as research to develop concepts for new, open, network-based collaboration tools, such as virtual situation rooms.
• Standards, interoperability, integration, and legacy. Four findings (5, 6, 7, and 8) are devoted to addressing the critical issues that affect system composability and interoperability. The findings call for research to identify design principles that can yield open standards such as communica-
Page 85
tions protocols, and research to develop generic technology that can facilitate interconnection and semantic interoperation of diverse information resources. Additional research is suggested to develop ways to predict the performance and reliability of the components used to construct software systems and to develop technological and architectural methods to maintain access to long-standing information and software assets while also enabling users to exploit new technologies as they become available.
• Adaptivity and reliability. Findings 9 and 10 deal with adaptation to uncertainty and change, including on-the-fly adaptation to changes in topology, load, or environment. These findings call for research to increase the adaptivity of networks and applications so they can function during or after crises and research to enable accurate assessments of the reliability of systems composed of potentially unreliable hardware, software, and people.
• Distributed systems performance. Finding 11 calls for research to improve understanding of how to reason about, measure, predict, and improve the performance of distributed systems, given that most crisis-support systems involve highly distributed configurations operating in unanticipated ways.