Page 25

3—
Information Technology Research Opportunities

Information Management

Workshop participants identified several areas in which improvements in information technology could have a significant influence on the management of information during each of the four different phases of crisis management (see Chapter 1). A fundamental goal of the use of information systems in crisis management is the ability to supply decision makers at all levels with the information they need when they need it. The information users who must be served during a crisis include, among others, crisis managers themselves, field workers, and victims or potential victims of a disaster. Supplying decision makers with information requires a number of capabilities. First, the appropriate data must be acquired, either as a preparedness activity or during the response to a crisis. Crisis responders require retrieval and access mechanisms to allow them to find and reduce to the essential items the information they need. Delivery mechanisms are needed to get appropriate information to the right people. Success in each phase of crisis management depends on successful information management in the preceding phase. For example, the response relies on the development of an effective mobilization database and plan in the preparedness phase, and integration of crisis response resources depends on the effective tracking of people and other resources during mobilization.



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 25
Page 25 3— Information Technology Research Opportunities Information Management Workshop participants identified several areas in which improvements in information technology could have a significant influence on the management of information during each of the four different phases of crisis management (see Chapter 1). A fundamental goal of the use of information systems in crisis management is the ability to supply decision makers at all levels with the information they need when they need it. The information users who must be served during a crisis include, among others, crisis managers themselves, field workers, and victims or potential victims of a disaster. Supplying decision makers with information requires a number of capabilities. First, the appropriate data must be acquired, either as a preparedness activity or during the response to a crisis. Crisis responders require retrieval and access mechanisms to allow them to find and reduce to the essential items the information they need. Delivery mechanisms are needed to get appropriate information to the right people. Success in each phase of crisis management depends on successful information management in the preceding phase. For example, the response relies on the development of an effective mobilization database and plan in the preparedness phase, and integration of crisis response resources depends on the effective tracking of people and other resources during mobilization.

OCR for page 25
Page 26 Information Acquisition Several research topics were described to improve acquisition of better information for use in responding to crises. One challenge is discrepancies between the data stored in crisis responders' geographical information system (GIS) databases and the ''ground truth." For example, crisis responders may have access to rough, outdated information on storage facilities of hazardous materials but may lack up-to-date, detailed information about what kind of material a particular building contains, even though such information is known to the operator of an industrial facility. New data management paradigms are needed that would permit geographically and administratively distributed GIS repositories to operate with one another in a more seamless and transparent fashion.1 Improving the collection of both input and response data during (rather than only after) a crisis is obviously important for keeping crisis responders informed during a crisis. In addition, mining the data after the event would facilitate formulating improved response plans for future crises by determining which response measures and mitigation efforts were effective. Such data sets would be invaluable in validating and improving the quality of crisis models. Integration and Interoperability Integration of information from a variety of sources and organizations is a fundamental issue facing crisis responders. Requirements for integrating data are not uniform—the requirements for speed, completeness, and quality of the information and the integration among organizations all vary depending on the phase and location of the crisis. Early in the response to a crisis, integration must proceed rapidly, often in an ad hoc fashion. Describing a California Department of Forestry incident management team that manages large incidents including fires, floods, and earthquake, Thomas O'Keefe observed at the workshop that these teams must be able to go anywhere at a half-hour's notice and must manage the up-to-several-thousand crisis responders arriving within 24 to 36 hours. Just getting this number of people to an incident quickly is a major challenge that represents only the first part of the problem. Integrating them for optimal performance is much more complex. Integration efforts must extend both vertically within an organization and horizontally among organizations—sometimes across a large number of organizations. Response to a major crisis in the United States, for in- 1The Open GIS Consortium, with participation from government and industry, is working to develop standards for such sharing of geographical data.

OCR for page 25
Page 27 stance, involves local, state, and federal governments as well as private-sector businesses and organizations. A crisis with international dimensions may be even more complex. Jack Harrald cited the recent crisis in Rwanda, in which hundreds of organizations and hundreds of thousands of people were involved. Indeed, at one point during the Rwanda crisis, each of more than 100 international organizations were producing information that in many cases was inconsistent. A number of significant nontechnical barriers impede implementation of solutions, including organizational resistance to sharing data or to interoperating, lack of overall system architectures, security constraints that make information sharing difficult, and both the absence of applicable standards and nonadherence to extant standards.2 In addition, workshop participants identified a number of ways in which research on interoperability and integration could make a significant contribution to improved crisis management. First, given the dynamic, rapidly developing nature of crises, improved techniques for the dynamic discovery of information relevant to a crisis and for the fusion of information from multiple sources are important. Central to discovery and fusion of information are techniques that help determine the accuracy, reliability, or "quality" of the information that is discovered and processed. Some key approaches to integrating information and facilitating interoperability of information systems involve the creation and management of metadata, the information that describes the format and content of other information, such as the fields in documents or annotations of video sequences. Standardization of the metadata describing the format of the many databases involved in crisis management systems could be achieved through agreement on XML DTDs (document type definitions, which are formal descriptions of what can appear in a document and how documents are structured),3 and more work should be done in this area. Metadata that describes the content or important topics covered by information objects and databases is more difficult to standardize but is a crucial part of integrating heterogeneous information resources. This type of 2Many other factors make it difficult to achieve interoperability. See Computer Science and Telecommunications Board, National Research Council. 1999. Realizing the Potential of C4I: Fundamental Challenges. National Academy Press, Washington. D.C., for a discussion of factors affecting all organizations as well as special challenges faced in a U.S. military context. 3XML is the eXtensible Markup Language being developed by the World Wide Web Consortium.

OCR for page 25
Page 28 metadata is often expressed using a predefined vocabulary, represented as a list of categories or more structured forms such as taxonomies and ontologies. Many taxonomies already exist in government agencies, and many others are being created. Developing technology to support the development and merging of taxonomies, as well as the application of these taxonomies to information objects, is an important research challenge. As part of these efforts ways must be found of coping with the evolution of metadata. Another challenge is to find ways to address heterogeneous standards, much as systems today must support multiple image or document format standards. Although metadata, ontologies, and the like are important tools for integrating data, further work is needed on approaches to both system interoperability (ensuring that systems can successfully exchange data) and semantic interoperability (allowing data arising from heterogeneous systems to be successfully interpreted), particularly when data integration must be conducted ad hoc and on the fly. The response to a crisis is characterized by the distributed generation of large amounts of unstructured, multimedia data that must be acquired, processed, integrated into the current situation model, and disseminated in real time to be useful to crisis responders. Technology that automatically captures the context of each piece of data would significantly increase its value. Even relatively simple techniques such as automatically geo-locating all input data elements would be beneficial. Techniques are needed to filter both incorrect and duplicative data items and to summarize and automatically convert unstructured data inputs into progressively more structured forms for subsequent analysis by both humans and models. Much of the unstructured data will be in the form of text, so research on text filtering, summarization, extraction, and event detection will be particularly relevant. Speech will be another important source of information, and exploiting it will require research on recognition in noisy environments, segmentation, and indexing. Video will become increasingly important, and techniques for video segmentation, summarization, and indexing will be required. Integrating and exploiting the rich information content in multiple video sources, such as might be obtained from crisis responders in the field, are additional challenges. Given that crisis management depends on the integration of information coming from multiple organizations and government agencies, each of which may have policy constraints regarding confidentiality, a challenge is to develop techniques that permit integration for the purpose of crisis management consistent with maintaining those constraints.

OCR for page 25
Page 29 Data Delivery One important element of data delivery is ensuring that data will be available when and where needed. Replication—the periodic copying and distribution of updated versions of database contents—is clearly a key component in availability. But simply replicating data in a safe location outside the region affected by a crisis is not necessarily effective if the crisis cuts all communication paths to the replicated data. One obvious solution is to increase the number of replicas, but determining what is an optimal design requires understanding the appropriate trade-offs both in the cost of providing and managing local storage and in the performance penalty entailed in keeping the replicas updated and consistent. Data delivery in a crisis situation is an example of the issue that David Maier pointed to in his discussion of limitations of today's database systems (see "Databases" in Chapter 2)—the need for database systems that include data staging and movement rather than just serving as repositories. The delivery of large amounts of information in real time or near-real time, for example, video or high-resolution satellite imagery, is a significant challenge, particularly when existing infrastructure has been damaged or delivering to mobile units is required. One option pointed to by workshop participants was the use of digital direct broadcast satellite services by crisis response teams deployed in the field. The Defense Advanced Research Project Agency's (DARPA's) battlefield awareness data dissemination project was cited as having developed technology that might be adaptable to the data delivery requirements of crisis response. Geographical Information System Performance An additional area of particular concern noted by workshop participants was the performance of GISs, which play an importantrole in crisis management. Participants noted that the major database system vendors such as Oracle, IBM, Informix, and NCR are rapidly improving both the functionality and scalability of the spatial capabilities of their standard database products. Over the next couple of years it is likely that the majority of large spatial data sets will reside in commercial database systems and not in specialized GISs. As managing terabyte-sized spatial data sets becomes as routine in the future as managing terabyte-sized commercial data sets is today, database vendors will need to provide a sufficient level of GIS performance. Information for People Like any other human-computer interfaces, crisis management systems should be developed using good user-centered design methods,

OCR for page 25
Page 30 including an early focus on users and their tasks; ongoing empirical measurement and evaluation of systems; iterative design and testing; and integrated focus on the end-to-end systems, which considers the larger social context in which they are deployed. However, crisis management poses a number of unique challenges (e.g., situations are nonroutine, rapidly changing, often very high risk, and so forth). Interfaces must be designed so that they can operate effectively in a high-stress environment, as well as be intuitive, given that they may be operated by users who have had only minimal training or who may not have operated a system since the last crisis—in a crisis, no one can effectively digest a thick user manual. Many of the most difficult human-technology interface issues are evident in the initial response stage of crisis management, but some occur in the recovery, mitigation, and preparedness phases as well. (In addition, as discussed below, there are important reasons to use the same tools and interfaces across all these phases.) The response phase is characterized by the need to observe, understand, and integrate a wide range of information sources; communicate and coordinate among the many different roles of and requirements for information; and, most important, make rapid decisions. The following general observations cut across many of the more specific comments below: • There is no Moore's law on human perception, attention, or cognitive and problem-solving capabilities. The scarce resource in human-computer interfaces is human attention. The situation is exacerbated in a crisis, because the novelty of the situation consumes attention that would otherwise be available. The issue of information overload is discussed further below. • Crises are nonroutine and complex. Thus they pose challenges for training and iterative design and evaluation of systems. In addition, humans resort to well-learned and practiced behaviors during times of crisis and stress. Creative problem-solving is difficult under these circumstances. • Communication and collaboration are critical. Crises involve a variety of phases, organizations, information needs, and roles for individuals. Often people who have not worked closely together are brought together in demanding circumstances. • Crisis responders are best able to make effective use of tools that they also use routinely. Priority should be given to developing tools and interfaces that are useful in both routine and crisis situations. For example, most users today are familiar with the Web browser user interface. • Decision making is key. People in all roles need to organize, abstract,

OCR for page 25
Page 31 and share information rapidly and flexibly in support of effective direction and coordination of crisis response activities. Presenting and Using Information Crisis responders need to get information quickly and flexibly. Today's interfaces often exacerbate this problem by relying on a limited set of input and output capabilities. Researchers should continue to push the hardware and software envelope to support new interaction styles (e.g., richer visualization, perceptual user interfaces, multimodal input, support for a range of motor and language capabilities) to eliminate this impedance mismatch. Advances are required at the cognitive level as well, where people's rich but fallible memories and vast amounts of general and domain-specific knowledge often do not match well with the information required by computer decision support systems. A richer range of interaction styles is also important to match the user's environment. For someone who is driving a vehicle, for example, an audio interface may be more appropriate than a screen display. At the forefront of all design improvements should be the goal of better leveraging and augmenting of natural human capabilities. Information presentation must be flexible. People need to extract relevant information rapidly, but which information is relevant varies across individuals and at different phases of crisis management. Not all the data collected during a crisis needs to be disseminated to all the various decision makers during the crisis (although capture of all the information may be invaluable later for analysis and training purposes). Some crisis responders need to get the "big picture," others need to abstract and integrate information, and others need access to finer and finer details. User interfaces that support integration of information with easily configurable "views" are needed. One approach might be to define a set of well-tailored products aimed at predefined crisis responder ''customer models" that then can also be fully customized by the user. Developing improved systems requires a better understanding of user requirements, information presentation techniques, information access strategies, and the development of flexible and modular architectures for information selection and presentation. Supporting Effective Communications and Coordination In crisis management situations, crisis responders are quickly brought together, both physically and virtually. Multiple existing infrastructures, bureaucracies, and individuals are quickly assembled into a virtual team. People need to quickly develop community and working relationships. A

OCR for page 25
Page 32 hybrid of centralized and decentralized approaches to controlling and managing information resources is required. Communications of all forms (one-to-one, one-to-many, many-to-one, and many-to-many) must be supported. Interactions will take place among crisis responders and between crisis responders and citizens. People need to speak the "same language" (or find ways of translating among languages)—a challenge that encompasses both multilingual issues (e.g., facility in 27 different languages was required in a recent California crisis) and semantic mismatches (which are more difficult to detect). Situational awareness provides an important background channel of activity (e.g., the command center "hubbub") and must be maintained in electronic environments. Situational action depends critically on having a good sense of the overall state of events. In addition, a common understanding of the general state of affairs gives all decision makers better information for making necessary trade-offs. Supporting Effective Real-Time Decision Making Under Uncertainty and Stress Crisis situations are characterized by rich, rapidly changing information flows and by tremendous uncertainty. People in stressful situations and conditions of information overload tend to resort to ineffective decision-making strategies. Simply providing access to information is not enough to support decision makers. Much more effective systems are required for helping crisis responders evaluate, filter, and integrate information. As one workshop participant put it, in a crisis, support must be provided to overwhelmed and distracted individuals who are, for example, less able to take on new projects or use new information technologies. This situation is different from more routine circumstances in which the traditional rule-based systems for decision support operate. Simulations and preparedness drills help in many ways, but training for unpredictable events is a difficult challenge. Current interfaces also do not typically provide support for quickly prioritizing tasks (though such tools have been developed in some fields such as medicine). Several measures can be taken to help mitigate these design challenges. Increased automation was one approach suggested to compensate for decreased capabilities and to reduce stress level. Mixed-initiative systems, which combine features of directly manipulable and agent-based systems, could also support more effective decision making in crisis situations. Systems could be asked to monitor important situations and highlight changes that signify problems. Systems could also suggest courses of actions based on the situation. For example, automatic triggers in a

OCR for page 25
Page 33 disaster where many people are displaced from their homes could suggest to an emergency manager what resources (e.g., blankets or cots) might be necessary. Better methods for users to interact with agents and more effective tools for automated reasoning would help. The utility of such systems will be enhanced if they incorporate emergency management current plans and the underlying assumptions. For example, if a plan has crisis responders entering a flood-stricken area to evacuate residents, a system designed to know planned entry and exit routes could provide an automated early warning if the plan becomes infeasible because a bridge along the route has failed. Another approach to coping with the needs of users under stress is to improve the ease of use of systems through a better understanding of the varied needs and capabilities of users. For example, systems that can monitor a user's performance would allow systems to adapt in real time to the changing capabilities of users under stress. Another area to explore is development of stress filters for the audio, visual, and textual (i.e., e-mail) information that is provided to crisis responders. For instance, one of the major problems during crisis response is the transfer of stress among the responders, a self-compounding problem akin to what happens in a noisy restaurant when customers speak louder in an attempt to make themselves heard. Systems that detect and defuse such stressful spirals would be useful Another issue important to effective decision making is understanding the uncertainty in presented information. Although people realize at an intellectual level that information, whether based on field reports or the output of a simulation, may be uncertain, there are few external aids to reinforce this. Displays tend to show crisp and clear boundaries, report numbers to several places of accuracy, and so on. Information presentation techniques that better represent the inherent uncertainty would facilitate a direct, intuitive, and more accurate understanding of the state of knowledge. Handling Information Overload The process of collecting, organizing, and disseminating information during the course of a crisis is time consuming. A single person in a command post may need to listen to and integrate information from dozens of people in the field to get a complete picture of what is happening during a disaster. Disaster situations force emergency managers to contend with 100 to 1,000 times the normal number of variables, and the impacts of this stress level should not be underestimated. The scale of information collection and dissemination during a significant emergency is vast. In a Santa Ana fire, for example, 15,000 radio transmissions might

OCR for page 25
Page 34 occur in a day. Person-to-person, verbal communications cannot cope with this sort of information traffic volume, and no one can keep track of the entire picture. Workshop participants suggested that innovations in speech recognition technology may be helpful in meeting this challenge. In addition, other computing and communications technologies are being developed that can help identify, retrieve, filter, prioritize, and integrate diverse sources of information to support a wide range of decision makers. Some users will need access to details and others will need a higher-level picture of a given situation. Systems that provide situational awareness and a shared view of the information can help in communicating with others. Overcoming Language and Other Barriers to Communication During routine and emergency incidents, 911 emergency operators, as well as firefighters, paramedics, and law enforcement officers, must deal with people who speak languages other than English. Language barriers are, of course, also a factor in military operations. Workshop participants pointed to the potential represented by the DARPA multilingual interview system developed for use in Bosnia (described in more detail in Box 3.1) in helping crisis responders communicate with citizens not fluent in English. Participants observed that this device's potential benefit for the delivery of public safety services could be large. Potentially, PC-based versions could be placed in every law enforcement, fire, and emergency medical dispatch center in the nation. A wearable version might be widely used by the fire service, paramedics, law enforcement agencies, correctional facilities, and hospitals. It would also be useful to build translator systems to cope with other communications barriers between crisis responders and citizens. Devices could be built to facilitate field communications with the hearing and speech impaired as well. Warning Citizens at Risk Early warning systems have been developed for many hazards, including earthquakes, tornadoes, nuclear plant accidents, and tsunamis. In the case of earthquakes, even a few seconds' warning can be useful. This was the case, for instance, following the Loma Prieta earthquake. Rescue workers were able to receive a few seconds' notice of an aftershock, giving them a chance to move to a safer location. Warning systems also play an important role in flash floods. In that case, there have been significant advances in detection but less progress in the dissemination of information from the detection systems and in the response to the warnings.

OCR for page 25
Page 35

BOX 3.1 Fielding a Multilingual Interview System One new technology developed by the Defense Advanced Research Projects Agency (DARPA) that has been rapidly translated into a fielded product is the Multilingual Interview System. The basic technology is simple, consisting of a speech recognition system that recognizes which of a set of 200 or 300 phrases has been uttered, consults an index, and pulls out a CD recording of that phrase spoken in another language. The system enables the user to get yes-or-no answers to basic questions, as well as directions using maps. Such devices must be designed to be portable and easy to update. To construct a prototype, DARPA packaged this system into a laptop device with a power supply and other components. The system has been provided in this form to users in the field in Bosnia, where it has aided In such activities as interviewing Bosnians about the locations of mine fields. A hand-held version is being developed by DARPA that could be placed in a coat pocket. Those who have used this technology have been very interested in its possible applications, and there appears to be a potentially strong commercial market, according to workshop participant Ronald Larsen. In all crises, providing up-to-date information to large segments of the public is important because it permits them to take appropriate actions, helps prevent panic, can speed remediation efforts, and can prevent follow-on crises. Widespread broadcasts are not necessarily the best approach—they can provide only limited situation-specific information and cannot provide details tailored to the needs of individuals, such as what evacuation route to use. Also, broadcast warnings are not well suited to disasters that have limited geographical impact. False alarms have the effect of decreasing the attention people give to warnings. New technologies like "call by location" and zoned alert broadcasts could help by providing more focused (and presumably more accurate) warnings, and more detailed advice on what actions to take. One approach identified as worthy of further investigation is what is known as a reverse 911 system, whereby the usual direction of interaction between citizens and emergency managers is reversed. In a crisis such as a fire or flash flood, such a system could automatically call all the households and businesses that might be affected, warn them of the impending danger, and instruct them on what evasive action should be taken. The approach can be extended beyond simply making calls over wireline telephones. For instance, the Federal Communications Commission has mandated that cell phone systems be capable of providing accurate information on location. One could envision exploiting this capability to include automatic dissemination of information to cell phone users

OCR for page 25
Page 37 tems as well as video and audio; tools to create and manage metadata; ontologies and indexing capabilities to support access and retrieval; delivery mechanisms to share previously captured experiences, including near-real-time availability to permit use of prior knowledge during the course of a crisis; and capabilities for adding commentary to captured information and conducting after-the-fact analysis. In addition, the capturing of such information in computational models—allowing computer systems to reason based on the collected knowledge—could be useful for both training and operational applications. In collecting audit trails, it is important to capture not only the actual state of events at a given time but also when and where information was received. For example, a fire might have jumped a firebreak at 0130, but this fact might not have been detected until 0150, and the information might not have reached the headquarters for the area until 0205. Because one of the goals of collecting audit trails is to test the effectiveness of different communication and information management strategies, it must be possible to reconstruct "flows" of information and determine where information bottlenecks or loss occurred or where delays or error were introduced. Workshop participants also noted that process and workflow techniques could be applied to the response phase of crisis management so as to find ways of capturing and representing "best practices." Of particular interest was finding a way of sharing these practices across administrative domains, which requires finding ways of translating organization-specific practices into practices that are applicable to a broader set of organizations. Another critical means of supporting crisis managers is to provide them with just-in-time training and help. The rudimentary technologies available today for providing such assistance, such as context-sensitive help mechanism, are inadequate, and research here would be helpful. Using Wearable Computing Participants identified some opportunities that development of low-cost, high-performance "wearable" systems incorporating GIS capabilities might offer. Such systems, which would be designed to be usable by untrained experts in the field, would operate using wireless communications (either terrestrially based or via satellite as such capabilities are deployed; see "Wireless Communications" in Chapter 2) as well as in a disconnected or only occasionally connected mode. They would need to be capable of storing, manipulating, and displaying both standard spatial features (e.g., roads and rivers) and transmitting and receiving real-time imagery, voice, and video.

OCR for page 25
Page 38 One can envision a firefighter's assistant for fighting forest and brush fires. In a wearable form with voice recognition software for hands-off operation, such a computer could provide critical pieces of information such as wind speed, fire boundaries, and temperatures in a visual display format. Equipped with high-precision Global Positioning System receivers, the wearable device could also provide responders with situational awareness about each firefighter. Workshop participants also suggested that wearable computers would be of considerable value in urban search-and-rescue operations, such as the operation mounted in response to the Oklahoma City bombing. Such tasks could be assisted by the advent of tools that allow crisis responders to use, update, and refine maps on the fly during a rescue. One might, for instance, provide each rescue worker with detailed information, such as building blueprints, on the environment they are working in. An important caveat is that search and rescue personnel may be reluctant to overly depend on information of uncertain accuracy provided by such a system when they are working in an unstable, dangerous building environment. Thus, wearable computers might be more imnmediately applied to capture of critical information from the field rather than delivery of information to rescuers. Information Infrastructure As discussed above, crisis management is an information and communication-intensive activity. Information infrastructure is key to all aspects of crisis management. In preparedness efforts, networks are used to provide training and conduct virtual exercises. In crisis response, networks support information interchange among crisis responders and the provision of warnings and other information to citizens and after the disaster strikes are used to register claims for disaster relief funds. The global disaster information network, GDIN,5 for example, is a concept for an activity to provide access to disaster information resources, produce integrated information products, and deliver information to decision makers. 5The Disaster Information Task Force, responding to a request from Vice President Gore, articulated the GDIN concept in its report Harnessing Information and Technology for Disaster Management—The Global Disaster Information Network (Disaster Information Task Force. 1997. Washington, D.C.). At the request of the transition team considering issues related to implementation of a disaster information network, the National Research Council issued a report on how such a network could best provide information for decision makers. See National Research Council. 1999. Reducing Disaster Losses Through Better Information. National Academy Press, Washington, D.C.

OCR for page 25
Page 39 Robustness Meeting the information requirements of crisis management depends on a communications infrastructure that is robust in the face of damage, particularly as greater reliance is placed on information technology to cope with crises and their aftermath. Whereas for some applications infrastructure can be brought in from outside the disaster area (as in mobile GIS systems or satellite terminals to provide commander centers with communications), in a number of other cases crisis response would benefit from a more survivable infrastructure. Two examples of such applications are tele-registration for disaster victims and the coordination of crisis response activities among a large number of actors (see Boxes 1.2 and 1.3 in Chapter 1). A second, related requirement for communications infrastructure is the ability to adapt to changing demands, manage traffic congestion, and permit priority overrides for emergency usage. In a crisis, loading characteristics may diverge from normal patterns and exceed normal loads—at just the time when large portions of the infrastructure may have suffered physical damage. These scaling and robustness questions arise in a number of large networks that are key to public safety, such as air traffic control; police, fire, and safety communications networks; and 911 and other emergency dispatch systems. These issues also arise in efforts to leverage the public Internet and private networks built using Internet technologies. Several networking research questions arise from these requirements. Networks that are self-adaptive, would, for example, be able to rapidly configure and assemble themselves as, for example, wireless infrastructure elements deployed in response to a crisis. Also, networks that can reconfigure themselves quickly in response to the effects of damage and changes in demand will be of much greater utility. Infrastructure that is able to degrade gracefully as components of the infrastructure are affected by a crisis would be less likely to completely fail in a crisis situation. By their nature, crises result in a change in the normal demands for communications. There is likely to be more traffic, as well as traffic of varying priorities. A research question that addresses this requirement is how to build networks that allow applications to interact with the infrastructure so as to allow the incorporation of capabilities such as priority override features or the recognition and management of information surges during a crisis. Also, because of the need to maximize a crisis responder's ability to utilize communications resources, it would be useful to develop interfaces that allow the combined use of both private and public infrastructure during a crisis, permitting crisis responders to exploit whatever infrastructure elements are available in the aftermath of a

OCR for page 25
Page 40 crisis. Finally, efforts cannot be directed solely at improving the infrastructure. Work needs to be done at the applications level to ensure that applications themselves are able to cope with less-than-optimal network performance. Applications intended for use in crisis situations cannot assume that large amounts of bandwidth will be available or that connectivity will be available on a consistent basis. Strategies for coping would include adapting the frequency of updates to the available bandwidth or falling back to activities that consume less bandwidth (e.g., transmitting text instead of multimedia data). Infrastructure for Citizens Workshop participants also noted that if the communications infrastructure is to be available widely, especially for use in interacting with individual citizens, then low-cost, ubiquitous access is required. A range of new capabilities might be enabled through the use of the Internet, particularly high-capacity, always-connected access (as opposed to low-speed, dial-up connections, which require an action to be taken by the user every time any Internet connection is desired). Participants noted that these always-connected, broadband services, such as those offered by cable modem and DSL technologies, are only available to a small fraction of the U.S. population today. Complex technical, economic, and policy issues surround the provision of broadband Internet access to residences,6 but deployment via a variety of technologies is proceeding, and it is useful to explore how such capabilities might be used for crisis management. In addition to enabling improved ability to interact with citizens, such as the opportunities discussed above for providing enhanced or focused warnings, deployment of these new high-capacity data services to the home offers some interesting opportunities for determining the impact of a disaster. As communications links to the home are upgraded to two-way technologies, with the deployment of two-way-capable cable systems or deployment of fiber to the curb, a large number of small, fully networkable devices will be deployed throughout populated areas. Interesting opportunities arise if one considers placing sensors in each of these boxes. For example, in a region prone to earthquakes, a cheap accelerometer (e.g., of the sort used to trigger air bags) might be included in each box. One possible use of such information is to validate and refine the predictions of damage models following an earthquake (see below). After an earthquake, in the areas badly affected, the sensors (or network nodes) 6The Computer Science and Telecommunications Board will be initiating a study of these issues in 1999.

OCR for page 25
Page 41 would cease to function, while those located outside that central region would provide indications of the shake intensity. Polling these devices after the quake could yield almost immediately, two sorts of information: a map of where the communications infrastructure had failed and a rough map of shake intensities in the surrounding areas. Modeling and Simulation Role of Modeling and Simulation Models are physical or mathematical representations of a system, entity, phenomenon, or process. Simulation is a method for implementing a model over time. Modeling and simulation can be applied to numerous phenomena—such as hurricane track and intensity, earthquake damage, and the airborne dispersion of chemicals following an accidental release—and they play important roles throughout crisis management activities.7 Some of these roles are listed below: • Planning. Models are used before disasters to help in planning. For example, the threat of a Hayward fault quake is troubling because it could be expected to have effects similar to those of the 1994 Kobe, Japan, quake, in which some 5,000 people died. The San Francisco Bay area traffic problems following a major quake would be very bad, and the area could be expected to be split and paralyzed. Insights gleaned from modeling and simulation can be used to establish traffic routing contingency plans for such disasters. • Mitigation. Models showing potential flood risk, for example, assist mitigation efforts by allowing the identification of economic incentives for implementing changes and by serving as tools for educating communities about the risks they face. • Prediction of damage before a disaster. The Federal Emergency Management Agency (FEMA), for example, makes use of predictive models of the paths of hurricanes, based on information from the National Hurricane Center, and plots outs factors such as potential damage to mobile homes and numbers of hospitals in the area, in order to make resources available in advance. Some models that predict specific amounts of dam- 7Modeling and simulation are important in many other domains as well. For an exploration of areas of common interest to the Department of Defense and the entertainment industry, see Computer Science and Telecommunications Board, National Research Council. 1997. Modeling and Simulation: Linking Entertainment and Defense. National Academy Press, Washington, D.C.

OCR for page 25
Page 42 age have not been as well validated—most are derived from Cold War era nuclear blast damage data—and thus are not relied upon much by crisis responders. • Initial damage estimates. After an earthquake, quickly and directly assessing the extent and distribution of damage is difficult because acquiring and synthesizing damage reports takes considerable time. The initial damage estimates are essential for directing response efforts, as well as estimating requests for federal aid following the disaster. The scale of this problem is illustrated by the Northridge quake, in which more than 3 million buildings in the Los Angeles area were at risk. With disasters of this scope, getting a clear view of the extent of the damage takes time. One type of tool used to assess quake damage rapidly is a model that indicates what a particular ''shake" means in terms of damage. The model, which includes building stock (structure type, age, etc.), critical facilities, and lifelines, as well as geological information and demographics, predicts the number of casualties and the need for shelter and hospitals across the various soil types. Research Opportunities Better simulation and modeling capabilities would enhance the capabilities of crisis managers throughout the phases of a crisis. Workshop participants identified a number of improvements in the design and use of models that could be helpful during a crisis: • Better meeting of the needs of crisis responders. The output from models is frequently not presented in a way suited to meeting the real-world information needs of crisis responders. Models frequently produce results in units not of interest to the crisis responder. For example, plume models of a chemical spill or release of radioactive material typically produce maps showing dispersion in parts per million as a function of time. What a crisis responder actually needs is something that automatically translates the concentration of materials into more easily interpretable categories such as "safe," "hazardous but not life threatening," or "life threatening" so that appropriate action can be taken quickly. Closely related is the need to allow nonexperts to use models. This requires, for example, that technical parameters of interest to experts tuning a model are separated from those of interest to decision makers. • Data collection to support the real-time use and validation of models. Models have limitations in their predictive capabilities. For example, even with all the background data used in earthquake models, being precise about damage estimates is difficult. In the area affected by the Northridge earthquake there were more than 100 8-inch water pipes, and

OCR for page 25
Page 43 yet a damage model of the quake would not be able to predict which specific pipes would be broken in the quake. Similarly, building damage models might provide results applicable to a class of buildings in an area but not to individual structures, and engineers would still need to be sent into the field after the earthquake to assess damage to individual homes and buildings. Incorporation of data in real-time data can significantly improve the output of models. Integration of data collected during a crisis would allow both validation of the results of a model against the actual situation resulting from the crisis and better prediction of the next stage of the crisis as it evolves. For example, as an earthquake mitigation measure one might deploy sensors in buildings that would provide data that could be combined with shake models to improve the accuracy of damage predictions following an earthquake. • Model interoperability. Models tend to be developed and used in isolation. To more fully exploit the results of models, techniques should be developed that better allow models to be accessed and integrated into information systems. In particular, it would be useful to facilitate such capabilities as the integration of real-time data with the results of models and the propagation of results and their uncertainty between different models. The value of such integration between data and models can be illustrated using the water-main-break example introduced above. Improved integration would permit crisis responders to enter data from field assessments indicating which particular pipes were in fact damaged and then to recompute a model of the water system to provide an estimate of water availability, something of obvious interest in planning firefighting activities. Also, with such capabilities in place, systems that would permit different crisis models to be plugged in could be built, allowing comprehensive and realistic characterization of a variety of different crisis situations. One application proposed during the course of workshop discussions was the use of simulations to enhance the realism of exercises. One specific aspect would be the incorporation of realistic levels of stress. Although, of course, a major research goal is to provide emergency managers with tools to better manage the large volumes of information (such as numerous reports from the field) or the coordination of a multitude of field activities, increased levels of stress are nonetheless an essential element of crisis management. Providing realistic stress levels, such as through the simulation of very high information traffic levels, would greatly enhance the realism and training value of exercises. Audit trail and similar information captured during crises could be used to enhance the realism of the pace and nature of simulated communications traffic.

OCR for page 25
Page 44 Electronic Commerce Electronic commerce and related technologies can play the role of both enabler of and impediment to effective crisis management. For the purpose of this discussion, electronic commerce (EC) technologies are defined broadly to include electronic means for obtaining information on the availability of physical goods, requesting goods, and paying for those goods; methods for entering and processing requests for benefits and paying those benefits; and computer security technologies needed to control the flow of information or protect information from unauthorized modification. In crisis response, the logistics function lends itself to the use of these electronic commerce techniques (even if the supplies are already in the possession of agencies responding to the crisis and if no additional funds will change hands). Problems Caused by the Increased Use of and Dependence on Electronic Commerce Although certain aspects of electronic commerce can be helpful in responding to a crisis, the routine dependence on electronic commerce can also serve to make recovery harder, unless the infrastructure supporting such commerce is able to survive the event that triggered the crisis. In fact, misplaced reliance on technology could itself trigger the crisis. For example, widespread power outages are considered crisis situations today, unlike years ago when there was no dependence on a power grid. In fact, it is this dependence combined with a concern about the reliability and correctness of our computer software that has led to concern for the year 2000 problem and its consequences—a potential crisis that many people seem to fear more than many natural disasters (see Appendix B). Some key EC technologies that are widely depended on include credit card authorization, which is itself often dependent on the telephone; automated teller machines; and computer networks. The principal issue raised in workshop discussions regarding dependence on electronic commerce was the need to ensure survivability of the critical infrastructure supporting EC.8 It is important to assess the reliability and survivability of different parts of the EC infrastructure in light of different kinds of crises, and, for those parts not likely to be available, one must not depend on them for recovery and must be prepared instead to mitigate the effects of their loss. 8See Computer Science and Telecommunications Board, National Research Council. 1999. Trust in Cyberspace. National Academy Press, Washington, D.C., for a discussion of critical information infrastructure issues and associated research topics.

OCR for page 25
Page 45 Benefits of Electronic Commerce in Crisis Management Discussion pointed to several possible benefits of EC technologies in crisis management: • Information available from EC systems. Some of the most compelling ideas involved the use of information normally maintained in EC systems to gauge preparedness, to locate available resources, and to reduce public anxiety and hoarding. For example, inventory information from suppliers might be accessed in real time to find the nearest availability for supplies. In anticipated crises (for example, from a hurricane where there is plenty of advance warning), point-of-sale data could be processed to determine preparedness—as indicated by the rate at which individuals are purchasing plastic sheeting, plywood, and so forth. Of course, this kind of access to supplier databases raises privacy, business, and independence concerns. • National emergency purchasing directory. Although ties to the local EC infrastructure can identify local availability of supplies, resources will likely need to be acquired from outside the crisis-affected area. An online purchasing directory with information that will enable finding suitable suppliers who also have available inventory can significantly speed up the shipment of needed supplies into the affected area. • Processing input from many sources. Besides the direct tie to EC databases, there is an indirect benefit of electronic commerce that can be exploited. In particular, many of the technologies developed and used extensively in electronic commerce can also be applied to other crisis management needs. The ability to process data from many sources, either through Web pages or though call centers located outside the affected area, can be particularly useful. If such benefits are to be realized during a crisis, relationships supporting such data exchange must be established in advance. Further, procedures must be put in place to credential emergency responders electronically so that the authority of the various players to make requests or to offer services can be determined. Pitfalls of Traditional Electronic Commerce in Crisis Management A critical aspect of the infrastructure for EC that will affect crisis management is management of the trust relationships between parties, such as the relationships between insurance companies and those insured, between citizens and relief agencies (citizens must trust that they are interacting with legitimate representatives of those agencies), and between

OCR for page 25
Page 46 contractors and suppliers (who need assurances that they will be paid for their services or products). One of the differences between EC as it is normally applied and EC during a crisis is that relationships will be more transient during a crisis. There will be new players, and it will be necessary to determine whether they are authorized players with whom one should do business. In traditional commerce there may be more time to build a trust relationship. In a crisis, one must decide quickly whether to honor a request. Perhaps the most significant pitfall of traditional EC systems with respect to their application in crisis situations is the rigidity of the rules regarding authorization of particular operations. This rigidity has been the best way to limit fraud in routine use of EC systems. Typically an organization has a single entity who is able to authorize purchases. In fact, for governmental agencies, this practice is often legislated. However, during a crisis, these rigid procedures could lead to delays or worse consequences. Some jurisdictions provide for delegated authority in particular situations, but today's systems for EC are not able to deal with this "conditional delegated authority." Workshop participants felt that one of the research goals for EC should be to provide for more flexible authorization policies that will maintain accountability yet support delegated authority and other exceptions. Such policies must be supported by the EC infrastructure as it is applied in the normal case, so that it will be available when needed. However, the conditional aspect of the policies will limit certain discretion from being applied except in the condition of a declared emergency. Research Opportunities Workshop participants considered the following to be key research opportunities for electronic commerce in crisis management: • Development of technologies and standards for escrow sites where citizens can store important information that they might need to access in a crisis but that might not be available if systems within the affected area are inaccessible. This escrowed data would include medical records, financial data, family contacts, and other essential records. The escrow technology must protect the user's privacy, while improving the survivability of the users personal information. • Determination of the range of authorization policies to be applied in emergency electronic commerce situations. One such policy to be considered is

OCR for page 25
Page 47 conditional delegated authority and mechanisms to allow other kinds of exceptions. • Development of a dynamic trust structure to support ad hoc or instant accreditation of participants with limited authority and limited powers of delegation. Such arrangements would greatly facilitate the purchase, for example, of critically needed supplies.