The Biometrics Myth

First, the good news

There are obvious attractions associated with biometric authentication systems. Apart from “looking the part’, users love them. Once enrolled, there is nothing to remember. Simply place the right finger on the reader, or face a camera, or perhaps speak into a microphone, and that’s authentication. Let’s face it, users are hardly likely to leave their fingerprint at home, and short of a transplant, cannot pass it onto someone else. From a management perspective, once the user is enrolled, there is little systems maintenance overhead. It is rare that a user needs to be re-enrolled, so password resets become a thing of the past.

So why a myth?

To illustrate the problems, let us look at one of the most popular biometric solutions – fingerprint recognition. The mechanism for capturing the print is not important to this discussion. It could be an optical reader, or proximity capacitance. It doesn’t matter how detailed the scan is either. All this affects is the magnitude of stored data. The important point is that once enrolled, fingerprint data is nothing more than a static pass-phrase. Whenever users try to authenticate themselves, their fingerprint is compared to the stored version – and a match opens the doors to the system.

There can be all kinds of (expensive) protection against “copying’ fingerprints. Photographs of prints can be detected by scanning for three dimensions, latex glove copies can be detected and even amputated fingers can be distinguished unless absolutely fresh! But all that is for nothing. Once registered with the system, the prints become a series of binary digits, just like everything else on the computer. There is nothing to prevent a hacker recording the “fingerprint’ as it goes up and down the wire, replaying a copy in a follow-on attack. And once one biometric system has been hacked, they are all exposed, as users will almost certainly employ the same finger for all system gateways.

Call out:
Once registered with the system, the prints become a series of binary digits, just like everything else on the computer. There is nothing to prevent a hacker recording the “fingerprint’ as it goes up and down the wire

There has to be a simpler way

There is a much simpler and considerably more cost-effective approach, employing encrypted “vaults’, and transition to such a system can be transparent to users. One day, they log into to their PC/domain as normal and in the background, the system builds an encrypted vault. From then on, users believe they are logging into their usual PC/domain – the look and feel remains unchanged.

What the user doesn’t see is that the PC/domain authentication is now being delivered automatically. The back-end system administrator can begin to force password changes, and the system negotiates a new password on the user’s behalf. The password could be made up from the maximum number of characters the password field will permit, and even be of non-keyboard characters (such as control characters etc). If the security administrator wishes to go further, then the system could even change the password on every login. At this point the static password system into the PC / domain really is set as securely as possible. The important thing is that while all this is going on, users are completely unaware of the new process. All delivered without user education.

The benefits, then, can be summarised:

  • ease of use (a single pass-phrase to remember) whilst enhancing domain and application security via a full static password management process.
  • the ability to generate complex dynamic pass-phrases that are used to log into the computer, network and applications. Not having to remember these dynamic pass-phrases enhances security whilst not requiring end user involvement.
  • easy to deploy, to use and to manage, without the need to introduce back-end processing or overheads.
  • considerable reduction in helpdesk incidents for password resets due to single pass-phrase implementation (typically 60% of all helpdesk incidents are password related).

How safe is vault architecture?

The answer is simple; as safe as you want it, limited only by how much you are willing to spend. In its simplest form, the user’s vault could be a pin-protected encrypted file that sits locally on the client, or on a network drive. The next stage is to store vaults off-line, in the form of USB memory sticks. The advantage of this approach is obvious should the computer system or laptop be stolen. Protection can be ratcheted up still further by keeping vaults in smart media, such as USB crypto tokens or crypto smart cards.

In summary, an encrypted vault system can bring significant benefits to an organisation in terms of optimisation and enforcement of strong static password processes and change regimes, delivered without impacting the user. In fact the user migrates from having to remember and manage multiple passwords to both PC /domain and application to having to remember just a single pin-phrase. The benefits of biometrics at one tenth of the cost.

Armoursoft Ltd. are exhibiting on Stand 680 at Infosecurity Europe 2004 which is Europe’s number one IT Security Exhibition. Now in its 9th year, the show features Europe’s most comprehensive FREE education programme, and over 200 exhibitors at the Grand Hall at Olympia from 27th to the 29th April 2004.