Officials can also build out a threat matrix for social networks that would rely on persistent automated searches, he said. For example, 10 people might be talking about a fight breaking out at the University of Phoenix Stadium, which could trigger an alert.
With video data, thousands of video streams can't efficiently be managed from a central command post, so tiny edge computers located near cameras can forward less bulky metadata to centralized servers. Software can then sort the metadata for facial recognition and license plate recognition.
Jensen couldn't confirm if that level of detail will be used extensively for the Final Four, but the capability exists.
"License plate recognition is an easier analytic than facial recognition," he said.
"The whole idea is to push analytics to as close to the network edge as possible, with the majority of the work done at the camera level," Jensen said. "If you were sending back all the video data it would add latency to the network and you'd need banks and banks of servers to collect and store it all."
Detecting unauthorized drones
Hitachi's software also potentially can be used at the Final Four and nearby to intercept radio controls for unauthorized drones, Jensen said. The technology exists to take control of that drone and move it to a safe location after focusing a video feed on the device to determine if it is a threat, he said.
"Some drones are very small and are used to take pictures, but we would be able to tell it if a drone has, for example, a payload of five pounds and a flight duration of 30 minutes, which means I need to worry about that," Jensen added.
To detect drones and their size, Hitachi has also relied on third-party technology that can analyze the radio spectrum signature being used by a particular flying drone. That information can be compared to a database to determine the drone's size and then evaluate its threat potential.
"Lately we've been concentrating on the commercial drone market because ISIS is going more toward commercial drones since they are cheap and have more payload capacity," Jensen said.
Analyzing data from many sources
Hitachi's software can cost from $100,000 to millions of dollars, depending on the size of the deployment. One city recently said it wants to use the software to monitor sensors on water supply pipes, to receive an alert if a pipe has been breached. A breach could indicate a terrorist attack to add a poisonous substance to a water supply.
"We can monitor data streams from almost any subsystem, from social media to radar systems to gunshot detectors," Jensen said.
Sign up for Computerworld eNewsletters.