Publications

For over fifteen years Sandia National Laboratories has been involved in laboratory testing of biometric identification devices. The key concept of biometric identification devices is the ability for the system to identify some unique aspect of the individual rather than some object a person may be carrying or some password they are required to know. Tests were conducted to verify manufacturer`s performance claims, to determine strengths/weaknesses of devices, and to determine devices that meet the US Department of energy`s needs. However, during recent field installation, significantly different performance was observed than was predicted by laboratory tests. Although most people using the device believed it operated adequately, the performance observed was over an order of magnitude worse than predicted. The search for reasons behind this gap between the predicted and the actual performance has revealed many possible contributing factors. As engineers, the most valuable lesson to be learned from this experience is the value of scientists and engineers with (1) common sense, (2) knowledge of human behavior, (3) the ability to observe the real world, and (4) the capability to realize the significant differences between controlled experiments and actual installations.

Over the last fifteen years, Sandia National Laboratories Security Systems and Technology Center, Department 5800, has been involved in several laboratory tests of various biometric identification devices. These laboratory tests were conducted to verify the manufacturer`s performance claims, to determine strengths and weaknesses of particular devices, and to evaluate which devices meet the US Department of Energy`s unique needs for high-security devices. However, during a recent field installation of one of these devices, significantly different performance was observed than had been predicted by these laboratory tests. This report documents the data analysis performed in the search for an explanation of these differences.