http://www.pbs.org/cringely/pulpit/pulpit20030515.html
Iris Recognition Could Mean the End of Physical Privacy By Robert X. Cringely Security and privacy always seem to be in the balance when we think about emerging digital technologies. Encryption vies with detection. Entire industries are built around the opposite ideas of maintaining privacy and invading it. Ultimately, we have to decide how much we are willing to give up in order to feel safe. Or at least, it seems we have that decision when actually we probably don't. Get ready for a brave new world where someone will likely know where you are at every second, whether it is at the movies with your kids or at the Bide-A-Wee Motel with your neighbor's wife. Sometimes, terrible events create business opportunity, and that's what happened to Steve Morton after the terrorist attacks on the Pentagon and New York's World Trade Center. Morton quickly repositioned his company, Connecticut-based Oxford Micro Devices Inc., entering the suddenly burgeoning field of homeland security. A fiercely independent engineer and entrepreneur trained at MIT and hardened by years at ITT, Steve Morton runs a fabless semiconductor company -- a company that designs and sells chips, but has other companies actually build them -- in bucolic Monroe, Connecticut. His chips are Digital Signal Processors (DSPs) and were developed over several years primarily through small Department of Defense contracts with no venture financing involved. The chips are devoted primarily to processing images and video, but Morton's initial commercial application was something completely different -- fingerprint recognition. And not just any fingerprint recognition: Steve Morton built fingerprint recognition into the trigger of a gun so it could be fired only by its owner or by any group of people whose fingerprints were programmed to be recognized by the gun. That product development decision, which on its face makes a lot of sense in a violent world where children sometimes play with guns, almost killed the company. It wasn't that the project was too ambitious: Morton built successful prototypes that could have been produced at a reasonable cost. But the gun makers wanted no part of it. "The fingerprint business was frustrating," Morton recalls. "Nobody would partner with us for liability reasons. There was just no support, and of course, the National Rifle Association hated it. People would see our demonstrations and predict we'd soon be rich, but it never happened. We got lots of good press, and I was on national TV several times, but we couldn't sell a thing." If you build it, sometimes they don't come. Before 9/11, Morton changed Oxford's focus from fingerprint recognition to physical security, building a system based on networked intelligent video surveillance cameras. It was a logical use for his chips and, as he says, "in physical security there is no equivalent to the NRA." Interest in his technology exploded after 9/11 and grew greater still following the Washington, DC sniper attacks. "The Washington, DC, attacks were a turning point," recalls Morton. "These weren't religious extremists crashing airliners into buildings, they were two guys with a $500 rifle wreaking havoc on the nation's capital. Everyone felt vulnerable and if they didn't like the idea of public security cameras before, now they don't worry so much about the privacy issue." The U.S. currently has about 10 million closed-circuit TV cameras used for surveillance and physical security. These are analog cameras built with 50 year-old technology, and the only way they are able to actively foil a crime is if someone happens to be watching the video from a particular camera at the exact moment a crime is taking place. The video tapes -- if they exist, since many cameras aren't recorded at all -- are useful at trial, but generally useless for actually stopping a crime as it is happening. The entire system is analog and no automation is possible on any level. It is archaic. Oxford's cameras (they are actually called Camputers) are different. Each has a digital sensor like the one in your new DV camcorder, and that sensor is connected to one of Oxford's A436 Video Processor chips. The network connection is Ethernet, and all video is compressed before it goes across the network and is eventually stored in that compressed state to save hard disk space on a central storage system. Turning to digital technology and adding lots of smarts to the cameras dramatically expands the capability of a video surveillance system. Now, instead of being limited to a few hundred feet of video cable, the cameras can be viewed over the Internet from anywhere on Earth. That means the rent-a-cop watching the screen can be in New York during the day and in India at night. It also means that all the compressed video can be stored from every camera practically forever since storage costs are constantly dropping. And the pictures can be much higher resolution, making it possible to read license numbers and faces where it might have been impossible before. Those are examples of using digital technology to do the same job better, but the real power of Morton's system (now being sold to airports and governments worldwide under the name "Boundless Security") is when the cameras, themselves, are intelligent. The cameras can, for example, be told to look for a particular face, a particular behavior, a sound (gunfire?), or even to look for a unique biometric measure like the iris of a criminal's eye. Iris patterns stabilize around the first year of age and don't change again. Morton's system can identify iris patterns through dark glasses or contact lenses and can do so almost instantly for thousands of people passing through airport metal detectors or subway turnstiles. Creepy, eh? Link together all the cameras in a country and even Ted Kaczynski (or Saddam Hussein) would have difficulty hiding for long. What's going on here that is so important is not just that the camera is smart enough to identify these things, but that every camera can do so and that, upon finally finding that wanted face in a crowd at JFK, it can announce to the network, "Hey, look at me!" With Boundless Security, the rent-a-cop role is played by the camera, itself. Response times get faster, there is less error, and fewer donuts are consumed. Response time, it turns out, is critical. "A year ago in Germany, a gunman killed 18 people in a school, most of them before the police even arrived," says Morton. "When a facility is attacked many people are killed in a few minutes, like at Columbine. The question is how can lives be saved before the cops come? Automating the process so the cameras can do the work means you can know instantly where the trouble is so you can tell people how to move away from danger. This is a huge advance. The danger zones can be plotted on a PDA or mobile phone. Even before the police arrive they can know what is happening where in the building. Errors are reduced and lives will inevitably be saved." Airports in Canada, the UK and Holland have already started using iris recognition for a variety of secure applications, ranging from customs clearance and immigration to secure border control to speed up travel time for passengers. Nineteen state prisons in the U.S. are iris-enabled for access control, and more recently, an iris recognition medical records management solution has been launched at the University of Alabama. In Peshawar, Pakistan, the UN Refugee Agency is using an iris-recognition system among Afghan refugees So this is not future technology, it is present technology. Right now, the U.S. physical security market is fragmented between mom and pop local companies and giant players like Bosch, Honeywell, and Tyco. The move to IP-based technology is inevitable, but with new tech fighting old tech and hybrid tech, it isn't always clear what to do. And with security data about to share the network with corporate operational data, suddenly the security and IT people have to start talking, which they have generally not done before. It is an inevitable culture clash. If you flood the corporate network with video, does that increase the vulnerability of that network? There are several such unanswered questions. What isn't going on here is the tracking of every individual by their iris patterns as illustrated in the Tom Cruise movie, “Minority Report.” The distinction here is critical. What Morton's Camputers can do is look at an Iris coming through a turnstile, and decide if it is the iris of some limited number of specific people on the lam. Once the iris pattern is detected it is compared to a dictionary of dozens or, at most, hundreds of fugitives. To do what was shown in the Cruise movie would require comparing the iris pattern against a database of every iris on Earth. Camputers can't do that -- at least not yet. What this isn't, of course, is an opportunity to return to the way things were or even the way we like to think that things were. The world has changed forever and with that change our approach as a culture to physical security has changed, too. Short of getting an eye transplant like Tom Cruise, there is no going back and the best we can probably hope for is that technology can help us feel a little safer. _______________________________________________ http://www.mccmedia.com/mailman/listinfo/brin-l
