Contributions to this post were provided by Executive Council Member Dan Lopresti, Michael Dunaway, Robin Murphy, and Nalini Venkatasubramanian.
Cell towers on wheels? Monitoring Twitter? These are just some ideas of how to monitor disasters and inform the public during an emergency situation. One of the Computing Community Consortium (CCC) sessions at the recent 2018 American Association for the Advancement of Science (AAAS) Annual Meeting in Austin, TX was on Rethinking Approaches to Disaster Management and Public Safety with Intelligent Infrastructure and these ideas were brought up during the discussion.
The session was moderated by CCC Executive Council Member Dan Lopresti, from Lehigh University, with participating speakers Michael Dunaway (University of Louisiana, Lafayette), Robin Murphy (Texas A&M University), and Nalini Venkatasubramanian (University of California, Irvine). They talked about the design and integration of intelligent infrastructure — including embedded sensors, the Internet of Things, advanced wireless information technologies, real-time data capture and analysis, and machine-learning-based decision support — which holds the potential to greatly enhance public safety, emergency management, disaster recovery, and overall community resilience, while addressing new and emerging threats to public safety and security.
With it being AAAS, the audience was made up of a wide range of disciplines. This was very evident during the question and answer session. The panelists were first asked about getting ahead of disaster relief and instead of being reactionary, using proper zoning in various protected environments (like marshlands). Dunaway responded that policies often lag behind, or in conflict with current science or changes in the environment. An example would be what we now know about the hazards of suppressing the natural cycle of wilderness fires—as we all learned through early exposure to Smokey the Bear campaigns—and the unintended consequence of the uncontrolled growth of underbrush that fuels wildland fires. We need to redesign systems or mitigation strategies, but those changes have to interface with the policy.
Another audience member talked about how disasters unfolded and how people need information at a specific moment but infrastructure is failing. How do we evolve the systems so that in the middle of a disaster proper information is shared? Venkatasubramanian responded that people do not know what to ask for in the middle of an emergency. Notification systems are one-size-fits-all, which does not really work (not geo-targeted, not actionable) and needs to be delivered over a broken infrastructure. She suggested that mixing in social hubs with the communications network is a way to make progress.
Finally, one audience member remarked that it is most important to get the right info to right people at right time, but fundamentally nothing has changed since 2005. Murphy responded and said the biggest thing that is lacking is that we as scientists don’t have a good domain model of emergency management. In a disaster, so many things are happening that may be working against each other. The solution is to conduct work domain analyses of the tasks and interdependencies. For example, the current approach of “fly a UAV, get lots of data, send it unfiltered to everyone, and with no provenance” doesn’t work. Instead, a work analysis shows that the amount of data often exceeds what can be transmitted over wireless networks, so agencies default to a courier system. This highlights the need for intelligent computing to snippet relevant data and route it to the agency that can use it. Venkatasubramanian responded that it is a lack of connectivity. We need to find a way to get all of the data that is being collected, through the cell towers on wheels (COWs) and by monitoring Twitter, analyze the data, and then turn it into actionable responses. Consistent connectivity of multiple data sources and its assimilation into actionable information in time remains an issue. The panel discussed the use of emerging edge computing technologies to process information at the device or within the network as a possibility to realize timely and accurate situational awareness.
For more information see the CCC@AAAS website. The CCC has already done a lot of work in this space. See a relevant white paper called Research Agenda in Intelligent Infrastructure to Enhance Disaster Management, Community Resilience and Public Safety as well as the Cricis: Critical Real-time Computing and Information Systems report from the 2012 Computing for Disaster Management Workshop. Also, see the Intelligent Infrastructure plenary and panel at the 2017 Computing Research Symposium.
