Researchers at the College of William and Mary in Virginia have developed a continuous authentication system for smart glasses devices that combines behavioral and voice biometrics.
As the IEEE reports, the system, called GlassGuard, analyzes six types of user interactions with their smart glasses devices, mostly revolving around how the swipe their fingers across an embedded touch interface. It also analyzes spoken commands from the wearer. For each individual metric, GlassGuard gives it a positive or negative designation, reflecting whether it seems to represent the authorized user or not. After four or five such user events have been analyzed, GlassGuard establishes an aggregated score to authenticate the user – and if it isn’t an authorized user, the device can be locked.
The researchers say that in initial testing with 32 users, GlassGuard achieved an accuracy of 99 percent with only a 0.5 percent false rejection rate. And they say it got these results after an average of only 3.5 user events.
It’s impressive so far, but given that smart glasses haven’t yet taken off in the mass market, it’s a system with relatively limited applicability. But its fundamental authentication elements represent areas of growing excitement in device security, particularly when it comes to continuous authentication based on behavioral biometrics; and so the researchers may find growing interest in their solution as they continue to test and refine it.
Source: IEEE Xplore Innovation Spotlight