Identify person by unique brain activity

“Brain-wave signatures, represented as the EEG signals of a person … are different from one individual to another, even when they perform the same thought or task”, “A brain-based biometric can be as strong as DNA-based biometric”

Touradj Ebrahimi

Brain of each human being is completely unique. Its structure is highly influenced not only by our DNA but also by everything we experience in our life. You can find people even with the same DNA – but life history is something that cannot be duplicated.

So brain activity is unique biometric every person has. And such biometrics are used in access control systems when security is needed. Simply, you can use your biometric as a password to gain access to resources protected from everyone except you. And brain activity has a lot of advantages over other biometrics traditionally used in access control systems (such as fingerprints). Here are the most important ones:

  • Brain activity is secure. In case of fingerprints, for example, we leave biometric in every place we touch with our hands, so everyone who needs to attack the system can collect & replicate it. Brain, in contrary, is safely hidden inside a skull.
  • Brain activity is changeable. You cannot influence other biometrics – your iris pattern, DNA, heart beats, fingerprints – are determined by nature and there is no easy way to modify them. If system based on these biometrics is compromised once, it is compromised forever. Brain activity, conversely, can be easily changed just by simple thought.

Brain activity can be captured by different methods. Most of them are quite expensive and require a lot of time and effort, and hence cannot be used in access control systems. EEG method is most extensively studied for person identification. Traditional devices for EEG recording are bulky & expensive too, but technology moves forward, and cheaper & more convenient devices are now developed.

EEG as potential biometric for person identification has been studied since 1998. First papers were published by Marios S. Poulos. References to these works can be found on Poulos’ website. He recorded background EEG – i.e. EEG recorded while people were resting – and used it to identify these people.

Field was also studied extensively by Ramaswamy Palaniappan. In his method, a simple picture is shown to user and then EEG is recorded during one second. Then this EEG is used to identify person. Method was tested on a huge dataset and its accuracy is more than 90%. Despite of such good performance, method is far from real-world application because it requires a lot of channels. Another potential drawback is that it based on gamma activity, which, accordingly to latest research, might be attributed to eye or muscle movements and not have anything to do with brain activity.

Pictures used to elicit gamma activity in Palaniappan's studies

On the whole, today there is a good pool of studies regarding person identification by EEG. These studies show that, in most cases, we can identify person by his or her EEG with a quite high accuracy. The only downside is that these studies include only off-line analysis of EEG data, without testing how proposed method will perform when embedded in real access-control system.

But in HUMABIO project developers went so far they did the real-world testing. HUMABIO is a project which aims at developing multimodal biometric authentication system – e.g. this system will consider different biometrics: face, voice, gate, EEG, ECG – to recognize user. As a part of this project, STARFAST – a Wireless Wearable EEG/ECG Biometric System – was developed. You have to put a special cap on your head and it will record EEG for one minute. After system will be able to tell if you are legal user or impostor with 79% accuracy.

ENOBIO EEG Authentication

Notice thar cap used to capture EEG is wireless and does not require special preparation (i.e. gel application) – just wear on and it works. Traditional equipment to record EEG is much more cumbersome. More info about equipment – on Starlab website.


Comments are closed.