An olfactory sensor filters the breath for biometric authentication


Similar to fingerprint and iris scans, biometric authentication is an essential factor in any spy movie, and trying to circumvent these security measures is often one of the main plot points. However, nowadays the technology is not limited to spies, as facial recognition and fingerprint verification are standard features on many phones.

Individual authentication based on breath smell using an artificial olfactory sensor may become possible in the near future, as this artist’s rendering shows. Image Credit: Institute of Chemistry and Materials Engineering, Kyushu University

Currently, scientists have created a new prospective odor option for the biometric security toolkit: human breath. In an article published in Chemical Communicationsscientists from Kyushu University’s Institute of Chemistry and Materials Engineering, in partnership with the University of Tokyo, have developed an olfactory sensor that can identify individuals by examining compounds in their breath .

Embedded in machine learning, this “artificial nose,” designed with a 16-channel sensor network, was able to authenticate up to 20 people with an average accuracy of over 97%.

In the age of information and technology, biometric authentication is an important way to protect valuable assets. From the typical suspects of fingerprints, voices, palm prints and faces to the less common options of finger veins and ear acoustics, there is a range of biometric data that devices can use to identify you. .

These techniques rely on the physical uniqueness of each individual, but they are not infallible. Physical characteristics can be copied or even compromised by injury. Recently, human scent has emerged as a new class of biometric authentication, essentially using your unique chemical makeup to confirm who you are.

Chaiyanut Jirayupat, Study First Author, Institute of Chemistry and Materials Engineering, Kyushu University

One of these was percutaneous gas – compounds produced by human skin. However, these approaches have their limitations because the skin does not create a high enough concentration of volatile compounds for the devices to detect.

Therefore, the researchers chose to test whether human breath could be used instead.

The concentration of volatile compounds in the skin can be as low as several parts per billion or trillion, while compounds exhaled through the breath can reach parts per million. In fact, human breath has already been used to determine if a person has cancer, diabetes, and even COVID-19.

Chaiyanut Jirayupat, Study First Author, Institute of Chemistry and Materials Engineering, Kyushu University

The researchers began by examining the subjects’ breath to learn which compounds could be evaluated for biometric authentication. A total of 28 compounds were defined as feasible options.

Based on this, they created an array of olfactory sensors with 16 channels, each of which can identify a particular range of compounds. The sensor data was then fed to a machine learning system to examine the composition of each individual’s breath and develop a profile that would be used to identify one individual from another.

After testing the device with breath samples from six people, the team found that it could recognize individuals with an average accuracy of 97.8%. This high level of precision remained stable even when the sample size was expanded to 20.

It was a diverse group of individuals of different ages, genders and nationalities. It is encouraging to see such precision at all levels.

Takeshi Yanagida, Study Leader, Institute of Chemistry and Materials Engineering, Kyushu University

Still, he says more work is needed before it can be integrated into the next smartphone.

“In this work, we asked our subjects to fast for six hours before the test”, concludes Yanagida. “We have developed a good baseline. The next step will be to refine this technique so that it works regardless of diet. Fortunately, our current study has shown that adding more sensors and collecting more data can overcome this obstacle.”

Journal reference:

Jirayupat, C., et al. (2022) Individual authentication based on respiratory odor by an artificial olfactory sensor system and machine learning. Chemical Communications.



Comments are closed.