Dang Nguyen

Investigating the possibility of applying EEG lossy compression to EEG-based user authentication

Using EEG signal as a new type of biometric in user authentication systems has been emerging as an interesting research topic. However, one of the major challenges is that a huge amount of EEG data that needs to be processed, transmitted and stored. The use of EEG compression is therefore becoming necessary. In this paper, we investigate the feasibility of using lossy compression to EEG data in EEG-based user authentication systems. Our experiments performed on a large scale of three EEG datasets indicate that using EEG lossy compression is feasible compared to using lossless one. Moreover, a threshold for information lost has been discovered and the system accuracy is unchanged if the threshold is lower than or equal 11%.

Wavelet transform and adaptive arithmetic coding techniques for EEG lossy compression

Electroencephalogram (EEG) has been widely used in diagnosing brain-related diseases, brain-computer interface applications, and user authentication and identification in security systems. Large EEG databases have been built and therefore, an effective EEG compression technique is necessary to reduce data for transmitting, processing and storing. In this paper, we propose an EEG lossy compression scheme in which EEG signals are undergoing a Wavelet Transform operation, followed by Quantisation and Thresholding, before being coded by Adaptive Arithmetic Coder. Our experiments are performed on a large set of EEG signals taken from two public databases and the results show that the proposed compression technique gives better performance than current techniques.

Investigating The Impact Of Epilepsy On EEG-based Cryptographic Key Generation Systems

Abstract In this study, we investigate the impact of epilepsy on EEG-based cryptographic key generation systems. Epilepsy is one of the brain disorders that involves in the EEG signal and hence it may have impact on the system. However, this issue has not been investigated. To solve this problem, we propose a system for cryptographic key generation from EEG signals, and experiment it with the Australian EEG dataset. We used parametric spectrum estimate technique for feature exaction, and devised an error-correction quantization technique that is useful for a noisy data such as EEG. We performed experiments on two groups of subjects, epileptic and non-epileptic to investigate the impact of epilepsy on the success rate of the system. Experimental results show that epilepsy actually has an impact on the performance of the system.

Emotional Influences on Cryptographic Key Generation Systems using EEG signals

Abstract This paper presents a research conducted to verify the influences of emotion on electroencephalogram (EEG)-based cryptographic key generation system. Emotion, such as negative and positive feelings, involves in EEG signal, and hence it may influence on the system. This issue has not been analyzed. Using parametric spectral estimation technique for feature extraction, and devised a quantization technique for error correction, a key is generated from EEG data. These techniques are believed to be suitable for a noisy data such as EEG. For experiment implementation, we use the Database for Emotion Analysis using Physiological Signals (DEAP) dataset. The emotional dimensions used are valence (positive and negative) and arousal (calm and excited) that is divided each in two classes: low and high. For experimental methodology, we performed on two groups of subjects, valence and arousal. Experimental results show that emotion actually has impacts on the performance of the system.

On the Study of Impacts of Brain Conditions on EEG-based Cryptographic Key Generation Systems

Abstract In this paper, we investigate impacts of brain conditions on an EEG-based cryptographic key generation system. Brain disorders, such as epilepsy and alcohol, involve in the EEG signal and hence it may have impacts on the system. This issue has not been analyzed in the literature. To solve this issue, we introduce a method for key generation from EEG signals, and implement experiments on the Australian EEG and the Alcoholism datasets. We use parametric spectrum estimate technique for feature exaction, and devised a error-correction quantization technique that is useful for a noisy data such as EEG. For experimental methodology, we perform on two groups of subjects, epileptic and non-epileptic, alcoholic and non-alcoholic to investigate the impact of brain conditions on the success rate of system. Experimental results show that both epilepsy and alcoholic actually have impacts on the performance of system.

Random Number Generators Basedon EEG Non-linear and ChaoticCharacteristics

Current electroencephalogram (EEG)-based methods in security have been mainly used for person authentication and identification purposes only. The non-linear and chaotic characteristics of EEG signal have not been taken into account. In this paper, we propose a new method that explores the use of these EEG characteristics in generating random numbers. EEG signal and its wavebands are transformed into bit sequences that are used as random number sequences or as seeds for pseudo-random number generators. EEG signal has the following advantages: 1) it is noisy, complex, chaotic and non-linear in nature, 2) it is very difficult to mimic because similar mental tasks are person dependent, and 3) it is almost impossible to steal because the brain activity is sensitive to the stress and the mood of the person and an aggressor cannot force the person to reproduce his/her mental pass-phrase. Our experiments were conducted on the four EEG datasets: AEEG, Alcoholism, DEAP and GrazA 2008. The randomness of the generated bit sequences was tested at a high level of significance by comprehensive battery of tests recommended by the National Institute of Standard and Technology (NIST) to verify the quality of random number generators, especially in cryptography application. Our experimental results showed high average success rates for all wavebands and the highest rate is 99.17% for the gamma band. Journal of Cyber Security, Vol. 6 3, 305–338. doi: 10.13052/jcsm2245-1439.634 This is an Open Access publication. c

EEG-Based Random Number Generators.

In this paper, we propose a new method that transforms electroencephalogram (EEG) signal and its wave bands into sequences of bits that can be used as a random number generator. The proposed method would be particularly useful to generate true random numbers or seeds for pseudo-random number generators. Our experiments were conducted on the EEG Alcoholism dataset and we tested the randomness using the statistical Test Suite recommended by the National Institute of Standard and Technology (NIST) for investigating the quality of random number generators, especially in cryptography application. Our experimental results show that the average success rate is (99.02%) for the gamma band.

On The Study of EEG-based Cryptographic Key Generation

Abstract Biometric-based cryptographic key generation is regarded as a data mining approach that uses knowledge discovery techniques to extract biometric information to generate cryptographic keys for protecting secured data by encryption. This application has been widely used in security systems to limit the weakness of passwords. Although conventional biometrics such as fingerprint, face, voice, and handwriting contain biometric information that is unique and repeatable for each individual, they are difficult to change to be used in different purposes. In this paper, we propose a system to exploit human electroencephalography (EEG) data as a new biometric for cryptographic key generation. This system provides high potential because EEG is impossible to be faked or compromised. Our method is evaluated using the EEG Alcoholism and GrazIIIa datasets, and is shown to reliably produce secure cryptographic keys with a 99% success rate.