Pitt Professors to Develop a Secure-CITI of Pittsburgh

PITTSBURGH—Oil spills. Landslides. Tornadoes. Plane crashes. Pile-ups. Floods. These are just a few of the emergencies and disasters that could occur—and have occurred—in Pittsburgh. When the call comes in, emergency management teams of Allegheny County respond. However, most calls come from bystanders who rarely can give the details needed to plan the correct response.

Researchers at the University of Pittsburgh are developing a system that will allow emergency managers in Allegheny County to make timely and better planned responses.

The system, a Secure Critical Information Technology Infrastructure (S-CITI) for Emergency Management, will integrate incoming real-time data from cameras and sensors and signal when data deviates from normal activity.

Funded by a $2.8 million grant from the National Science Foundation, S-CITI will be the first of its kind to simultaneously integrate data from multiple sources, and be applicable to all cities. Cities such as Los Angeles have earthquake detectors connected to emergency managers, but no city has a comprehensive system where data from utilities, the National Weather Service, and traffic sensors are integrated in real time.

"Currently, emergency information is correlated in someone's head. We want to make it much more automated and efficient," said Daniel Mossé, associate professor in Pitt's Department of Computer Science and principal investigator for S-CITI.

Already deployed electricity, gas, water, and temperature sensors as well as traffic cameras will be utilized, and new sensors will be added. For example, stationary cameras that monitor traffic could be replaced with rotating cameras that take pictures in the case of a landslide.

Mossé and his team are working closely with the Chief of Allegheny County's Department of Emergency Services, Robert Full, who has endorsed the project. Talks are underway to evaluate the needs of the county's emergency managers, identify security concerns, and discuss how S-CITI can best accommodate the county.

"The computer scientists will identify the most appropriate ways to represent the information, integrate it, and merge it without revealing anyone's secrets, but still be able to develop coordinated action for multiple organizations in emergency situations," said Louise Comfort, coprincipal investigator and professor in Pitt's Graduate School for Public and International Affairs.

Security is quintessentially important, said Comfort. Allegheny County has 130 municipalities, each with its own secure information. In order for a central system to be implemented, each jurisdiction must consent voluntarily to release its information. Mossé envisions a system that uses the information without releasing it, except in an emergency when the data is needed to coordinate a response.

"Computers can marshall all kinds of complex and diverse information, which can be presented to the decision-makers in a clear, logical, and timely way. This will assist the emergency managers enormously," said Comfort. "They need to be able to review the information in a way that's easily comprehensible, then make a decision."

A critical piece of the system will be a learning module, which will analyze postemergency data and use the results for future pre-emergency planning. For example, explained Mossé, if one day the system detects much higher than normal usage of gas and water and lower than normal usage of electricity, it will alert emergency managers.

"Maybe there was an earthquake or a landslide and some electricity poles were downed and some water pipes and gas lines were broken," said Mossé. "However, the emergency managers may say this is a normal situation on Thursday afternoons because people barbeque and fill up their pools, and they turn off the lights in their house because they're outside."

If the change in sensor activity is not an emergency situation, the emergency managers will be able to program the system to ignore the signals on Thursday afternoon. The system will learn to alert the mangers only in actual emergency situations, saving time and labor.

The system could be useful in detecting such potential security threats as harmful materials entering the city on trucks coming through the tunnels. It also could alert emergency managers when information is not coming in that should be coming in, such as when power lines go down.

"We're beginning to be able to address very complex problems that we couldn't imagine addressing before because it would have been too hard," said Comfort.

Mossé's team will spend the first few years of the S-CITI project building a prototype of the system within the Department of Computer Science and then within part of the University. Potentially, the system will have direct links with the Pitt Police. When the system is sufficient, it will be deployed in parts of the city, and Mossé and Comfort eventually would like to see it used extensively in Allegheny County.

Rami Melhem, chair of Pitt's Department of Computer Science is coprincipal investigator, and other members of the team include department faculty members Ahmed Amer, José Carlos Brustoloni, Panos Chrysanthis, Milos Hauskrecht, Alexandros Labrinidis, and Kirk Pruhs.

The NSF Information Technology Research (ITR) program reviewed more than 1,000 proposals this year for the medium-level award and funded approximately 20 percent of the proposed projects. The ITR program funds innovative multidisciplinary research that extends the frontiers of information technology, leads to new and unanticipated technologies, creates revolutionary applications, or provides alternative approaches to complete important activities.

###

12/5/03/tmw