Currenty physical traits which can be used as identifiers include facial shape, fingerprints, iris patterns, retina patterns, hand geometry, speech, handwriting / keystroke order and finally even wrist vein patterns. For these identifiers to be physically compromised it would involve a very gory operation...

It is the task of biometric products to apply a mathematical algorithm to the identifier (retina, or fingerprint for example) to determine if a user is a who they are pretending to be. In devices where a finger print is the means of identification, a small device will scan the fingerprint of the user and momentarily record the fine details. Where fingerprints are concern this may include features such as the finger prints' hills and valleys, the direction and branching points, line endings and dozens of other minutiae.

The fingers features can be acquired optically, thermally based on temperature differences between the fingertip's valleys and ridges, by a pressure sensor or via a capacitive sensor which is essentially a small silicon chip with many thousands of sensing elements. For instance Infineon's "FingerTIP sensor uses the capacity difference between the surface of the sensor and the surface of the finger. The capacity measured at a ridge differs from that at a valley. Thus, about 65.000 capacitors acquire the data in a field of 224 x 288 pixels and transform it into a digital signal."

Optical sensors, consisting of a lens and/or prism optical system added to a CCD or CMOS camera, measure light deflections produced by a finger on a sensor surface. Thermal sensors rely on measuring temperature differences at the fingertip so they can be influenced by changes in temperature in the surrounding environment. Generally thermal sensors must operate in a O° C or above ambient temperature.

Different means of storing the identifiers exist, and while their are benefits to each method there is no absolute best. Some companies opt to store all identifiers on a central server, where user information is sent during authentication. Other technologies store the identifier on the local hard drive in a highly encrypted format. To minimize the ability of an individual going around the biometric security features some companies even enable users of their devices to add special protective layers to the hard drive. In these configurations the file allocation table (FAT table) is an integral part of the security system protecting their computer. When the OS attempts to access the FAT table the software side of the biometric security system can prompt for an authentication of the user. If the data doesn't match the hard drive remains physically locked out at the most basic level, rendering all information stored on it unusable.

Restricted access is good, but what if you are rejected during the authentication procedure from accessing your own computer? It can happen if you have a large cut on your finger for instance. In such situations you would be in the same position as if you had forgotten your password. What would happen if you were just rejected for no reason, or even worse some one else had a finger print similar to yours and was able to gain access your computer? Well the odds are statistically in your favor that neither of these situations will occur. Currently, some biometric devices boast False Acceptance Rates (FAR) of as low as 1 in 1 billion, while the industry average sits at about 1 in 10,000. Conversely the False Recognition Rate (FRR) is typically about 2%.

Of all the identifiers in the realm of biometrics the most accurate and secure are the retinal and iris scanning technologies. As your eyes are generally speaking very well protected and extremely unique there is much less room for errors to stand between you and your data. Though on the other hand, if you experience any kind of physical trauma to your eye you might be out of luck...

All in all biometrics is the final piece in the security triangle and the most secure. With the growing number of networked computers, and volume of sensitive data being stored on those machines the need for a means of securing data is ever increasing. Biometrics still seem to be marginalized into the James Bond category as far as the mainstream consumer is concerned. For instances where security is paramount, it offers a level of reassurance that is not there with passwords or token security devices. Think of it this way. If you run a website and your server is stored with hundreds of other servers at the hosting companies building, would you feel more secure knowing a retina scanner or hand geometry security device was protecting your server from tampering, or lock and key?