Body odor passes smell test as biometric

Researchers at the Polytechnic University of Madrid are exploring a new form of biometric authentication - body odor.

Turns out your smell - which may be just as offensive as the next guy's - is at least unique.

To prove that point, one only need to think of bloodhounds that can track people by following their scent. The researchers, however, admit that the technology they are experimenting with is not on par with traditional four-legged sensors.

The Polytechnic University research group known as Group of Biometrics, Biosignals and Security (GB2S) is working with the Spanish engineering consulting firm Ilía Systems Ltd. on this biometric study of "personal odor."

It turns out there are recognizable patterns of each person's body odor that remain steady, the researchers found. In addition, the accuracy rate of identifying a person by their unique odor turned out to be higher than 85%. Those numbers remain constant, the researchers say, even as body odor varies due to disease, diet change or even mood swings.

As part of the project, Ilía Systems has developed a sensor that can detect volatile elements present in body odor.

Still, the accuracy percentage puts odor on par with current biometric identifiers, which are often criticized in the security field and by end-users for unacceptable accuracy rates. While multimodal biometric measurements are improving the accuracy rates of the technology, some say the overall security and usage of biometrics is disappointing.

But researchers say capturing body odor can be as easy as someone walking past a sensor and would be less intrusive than fingerprint readers or iris scanners. The researchers envision the sensors being used at security checkpoints at airports or border crossings.

Of course, the ease of data collection would certainly raise eyebrows with privacy advocates.

The GB2S group is no stranger to biometric innovation. It won an award at the recent ActuaUPM competition for its biometric application for mobile devices, called BiomMo.

The application focused on identifying a person who performs a series of gestures while holding their accelerometer-equipped mobile device, and by having the user take a picture of one or both of their hands for analysis of finger and hand shape.