Various answers . . . .
I'm not a technician, so what follows is a lay understanding of these systems.
"On/off switch". It's my phrase originally, but it would be more accurate to describe these devices as biometric locks for the computer. The application for the biometric reader is to determine whether the finger presented at sign-on matches the finger that was enrolled. This is something less than an ATM bio-ID system, because there the security system has to match the presented finger to a file of all users. But both are closed systems, entirely contained either (a) at the desktop [simplest form in the Compaq example] or (b) in a proprietary network [such as Windows NT or a bank system]. (The Mastercard/IDT smart card arrangement is analogous here, as well.)
In a closed system, the security and privacy demands are minimized. There is no major concern about outside hacking of biometric encoding (although you could have an inside job, certainly, if anyone had the skills), since there is no access to outsiders. Same with privacy -- as long as you trust the employer or the bank not to sell your data elsewhere, which seems a pretty safe bet, your privacy won't be compromised. Further, the demands on the algorithm for accuracy are also minimized in a closed, and therefore small, system.
Under these circumstances, the Identicator software using an unencrypted, minutiae-based algorithm is about as valuable as the Identix encrypted, pattern-based algorithm available for much higher prices. Under different circumstances, though, it is a much different story.
The Compaq/IDT arrangement has very limited usefulness. It helps sign-on security at the desktop, and it can cut down on "lost password" problems at the local administrator level. But that's about it. From an Identix point of view, though, it does serve the purpose of introducing a biometric device, however limited, to public consideration.
Identix is working on much more ambitious applications. These will result in access to remote databases over telephone lines, with the presentation of biometric identification for access at the interface between the network and the inquiring digital certificate. Once you are out there in the open with your bio-ID, you will need encryption, a more secure algorithm, and an identifier incompatible with law enforcement databases (i.e., non-minutiae). You are also much more likely to be to be one of many, many more potential entrants into the data base, so you will need a more descriptive (more detailed) bio-ID. This will also ultimately be true in the smart card arena, when these portable bio-IDs get plugged into desktop readers or POS devices nationally.
Identix can hold the encrypted bio-ID file on its desktop device (not the computer itself, unlike the Compaq arrangement), or in a network server. Either place is likely to have better security than the hard drive of a PC. I don't know how someone would get into a Touchlock. Network servers used for this purpose should have firewalls, etc. Individual PCs are more hackable, although it will still be a pain for someone to try to do this one computer at a time. The real problem that the Compaq/IDT system doesn't appear to address (and doesn't need to address at this preliminary stage of biometric infrastructure buildout), is the potential transmission "in the clear" of a fingerprint file across phone lines. That application, I think, is well beyond the reach of this $99 product -- and well beyond the reach of the Mastercard/IDT pilot. But it is what Identix is doing with Oracle.
Although you didn't ask about this, APIs come into play when files are held at the network level. As you have noted on the NRID thread, it is up to the biometric provider to decide which API to use. Fowler indicated in the conference call that IDX would wait to see which API won out, and then conform to that one. My guess is it won't be HA-API. |
