In this modern age of digital this and distributed that, the average man on the street is up to his ears in information, so much so that its hard for him to know how to filter the information to find what is most useful to him. Its even worse for powerful corporations or institutions which have the resources to actively seek out even more information. Take for instance the UK's insistence on putting surveillance cameras everywhere. As of early 2011, there were estimated to be approximately 1.85 million of the digital eyes watching the country's roads, sidewalks, and alleys. In the face of that much digital information the question becomes - how can you even begin to process it all
The ability to process huge amounts of surveillance data in real-time is an important question to answer. Currently, most surveillance systems are designed to simply keep a record of what is going on so that people can go back later and see what happened after the fact. But Professor Umakishore Ramachandran at the Georgia Tech College of Computing and the director of the Star Center thinks that these surveillance resources can be better utilized to create what he calls situational awareness. His idea is to use technology to be able to identify threats and problems in real-time to create safer and more secure areas. He used the simple example of the Atlanta airport, saying that there are over 500 wired cameras currently installed there, and that the administration has plans to install thousands more. "All of these cameras come to a monitoring room, and there's a human sitting at a bank of 20 monitors," Professor Ramachandran says, "and he or she is using a joystick to move around and see what is happening. It doesn't scale well." He says that a technological solution is the best way to go in order to deal with this problem of too much data.
Digital Situation Awareness
He calls it situation awareness. His idea is to combine a number of different technological tools in order to be able to correctly identify potential security risks. There are a variety of technologies available which can be applied to identifying threatening situations, but each one of them has weaknesses. He mentioned computer vision as one possibility, although he was quick to say that computer vision algorithms do not work great under all conditions. He also suggested monitoring the movements of authorized personnel in order to detect unusual movements, because many security violations are actually committed by employees. But most importantly, he emphasized combining several different sensory styles to work together in order to identify security threats in real-time, because only by looking at each situation through a variety of modalities does a system have a good chance of doing that.
There are a number of ways such a system could go wrong. If a system is too touchy, identifying too many potential security problems, false positives, can cause the human security personnel to stop treating the system seriously. It is like the story of the boy who cried wolf too many times, except that in this case the boy would be a distributed system of thousands of cameras and RFID sensors. But boys and computers can be equally dismissed as irrelevant by annoyed adults. Also, a system which gives false negatives - fails to detect an actual security threat - is worse than useless because it would let down the people who are depending on it to identify potential security threats.
Science fiction or reality
Professor Ramachandran gave some simple examples to illustrate how such a situational awareness system would work. "Lets say that everybody has a boarding pass. So on the boarding pass, suppose I put an RFID tag. One of the common types of security violations that passengers do in an airport is that somebody checks in some baggage but does not board the flight. If you check in baggage but do not board the flight, that is a security violation. The simple solution is that you put the same RFID tag on the boarding pass and on the bags that you check in. And now you've checked in your bags, and then you go to a bar and have a drink, and for some reason you forget your departure time. In that case the security personnel can know that bags have been checked in but their owner has not boarded the plane. Further, if there is an RFID network in the airport, the security personnel can find out where the RFID-tagged boarding pass is located in the airport, and they can use cameras to look at the location. And somebody can go and catch that guy." That is not just a theoretical example, that is the illustrative example that Professor Ramachandran set up in his laboratory.
Other security threats that can be detected with a technological sensor network would be loud noises, or sudden shifts in movement, or people in unauthorized areas. Anything that is anomalous is something that can easily be detected through a sensor network.
Another use for a situation awareness network would be in a flash flood, such as what happened last month in Seoul, or the flooding in New Orleans during hurricane Katrina. If the quickly-rising flood waters could have been detected immediately, then steps could have been taken to adjust drainage areas and turn on pumps in order to avoid the rising water. Such a system could have saved lives.
Here and there and everywhere
Having huge networks of cameras, RFID networks, water and motion sensors, and everything else that can be tied together to create situation awareness network can prevent disasters and save lives, Professor Ramachandran said. But of course, there are also costs in making such networks. The professor suggested that cities or companies that stand to benefit from such networks can put them together. The individual sensor nodes are not that expensive, but the hard part is making it all work together in a large scale. That is what Professor Ramachandran's research focuses on, turning large-scale sensor networks into real-time threat-assessment machines.