Khazaimatol S Subari

Analysis of effective channel placement for an EEG-based biometric system

This paper discusses the potential of the EEG signal for implementation of a practical biometric system using 4 or less channels of 2 different types of EEG recordings. Studies have shown that the EEG signal has biometric potential because the signal varies from person to person and is impossible to replicate and steal. Data were collected from 10 male subjects while resting with eyes open and eyes closed in 5 separate sessions conducted over a course of 2 weeks. Features were extracted using the autoregressive (AR) model and analyzed to obtain the feature set. Results show that data from eyes open and eyes closed using 4 channels gave good classification rates of 96% and 97% respectively and that data recorded from 2 channels gave classification rates from 90% to 95%. Classification rates from 1 channel ranged from 70% to 87%. The average time taken for recognition was 0.38 seconds at the point of recognition. Based on these results, there is potential for implementation of an EEG-based biometric system.

Effects of diseased ECG on the robustness of ECG biometric systems

This paper looks into the effects of diseased subjects on the recognition rate of an ECG biometric system. A novel technique for feature extraction, linear predictive coding, is implemented along with neural networks for the classifier. Diseased ECG has been shown reduce the recognition rate of the system by only less than 1% and thus the system is robust towards diseased ECG. This allows for the system incorporating linear predictive coding to be used in practical situations where some users may not be aware of their health state and may have diseased ECG signals.

QRS-complex of ECG-based biometrics in a two-level classifier

This research is based on an ECG biometrics system which segments the QRS-complex, extracts the non-fiducial features and sends the data to a two-level classifier. For spectral analysis, the discrete Fourier transform (DFT), and discrete cosine transform (DCT) were used to transform the signal, before principal component analysis (PCA) is used to reduce the feature vectors. From here, statistical parameters were computed for the classifier, where the first level is denoted called feature matching (FM) and the second level is the Neural Networks algorithm (NN). The system is tested on two databases. Database I consists of 45 subjects with 10 recordings each (recorded on the same day) while Database II consists of 35 subjects with 20 recordings each (recorded on separate days). The accuracy measures were is 99.176% and 96.67% respectively.

Simulation of frequency modulated continuous wave ground penetrating radar using Advanced Design System (ADS)

This paper explains about how a frequency modulated continuous wave (FMCW) standard system in Advanced Design System (ADS) software is implemented for ground penetrating radar (GPR) experiments using the built-in Design Guide as reference. At the moment, there is no model created especially for GPR investigation in ADS. Therefore, the purpose of this paper is to give an idea on how GPR experiment results can be simulated and analyzed without having to perform the actual experiment. Simulations have been done in ADS to see how the components change affects the baseband spectrum. The final circuit system was designed based on the spectrum line stability.

Design of bow tie antenna in CST studio suite below 2GHz for ground penetrating radar applications

This paper explains how a bowtie antenna is to be designed for ground penetrating radar (GPR) applications. The considerations to be taken into account include to design the smallest antenna possible for low frequency range (below 2 GHz). This is because the test equipment used, Vector Network Analyzer by Agilent Technologies can only work up to 3 GHz. Bowtie antenna is chosen because it has better directivity compared to other antennas such as horn, and it is easier to fabricate using microstrip boards which are available in the university. At the end of the design, the operating frequency range is determined and the design is fabricated to be used for measuring samples like soil and concrete.

Comparison of speaker normalization techniques for classification of emotionally disturbed subjects based on voice

When reviewing his clinical experience in treating suicidal patients, one of the authors observed that successful predictions of suicidality were often based on the patients voice independent of content. Research has shown that the Gaussian mixture model of the mel-cepstral features of speech can be used to distinguish the speech of suicidal persons from that of depressed and control persons with high classification rates. Since the vocal tract length vary from person to person, can the classification rates of suicidal persons be improved through speaker normalization? We approach this problem by warping the frequency axis of the mel-cepstral features. The results show that two different approaches yielded the best results: i) by using the maximum-likelihood approach in a gender-independent database to compute the warping factor for a nonlinear warp and ii) by a transformation of the first three formants in a gender-dependent database to compute the warping factor for a linear warp.

EEG and emotions: α-peak frequency as a quantifier for happiness

This research investigates the relationship between the electroencephalography (EEG) signal and basic human emotions. In the first experiment, EEG signals were collected from electrodes at locations Fp1, Fp2, F3 and F4 from 4 male and 4 female test subjects while exposed to audio-visual stimuli. The stimuli were selected to evoke 4 groups of emotions i.e., sad, fear, happiness and disgust. The signals were then processed to remove artifacts using a novel modified empirical mode decomposition (EMD) technique. Subsequently, spectral features derived from the α- and β-bands were derived from the artifact-free EEG using a modified Welch periodogram technique with emphasis on finding the optimum number of Welch segments and epoch length. The hypothesis derived from the first experiment was subsequently tested on an additional 7 male subjects. It was observed that the α-peak frequency consistently had the highest magnitudes for happiness-evoked emotion for male subjects. Based on this observation, we speculate that the α-peak frequency can be used to quantify the level of happiness experienced by an individual.

A Novel of Surface Reflection Method Using Radio-Wave for Soil Density Estimation

Soil density is one of the important parameters to be investigated in civil, geological and agricultural works. Unfortunately, the challenging issue is found on a suitable model in determining accurately the soil density. In this article, a new soil density model based on radio-wave surface reflection method is presented. The development of the model is based on result analysis collected from the experiment. Then, comparisons with related theoretical models, Hallikainen and Topp, are performed. The experiment is performed by using a vector network analyzer (VNA) that generates radar signal and recording return loss (S11) from a horn antenna. In the analysis, two new proposed soil density models have shown good agreement for soil density from 1.1 g/cm 3 to 1.7 g/cm 3 for sand and silty sand samples. This is verified when the model able to predict real samples as the one used in the experiment and result shows a very small relative error within 0.05% and 6.87%. Additionally, spectrograms in real time are produced in this study in order to observe more on the soil density. By using the proposed developed models, soil density estimation can be easily determined with minimal data input such as soil type, return loss and reflection coefficient by using regular radio-wave devices.

Soil density prediction tool using microwave ground penetrating radar

In this paper, a microwave surface reflection method is proposed to analyze the effect of soil density with its electrical properties using ground penetrating radar (GPR) principal. Three types of soil samples are chosen for the analysis of this project, namely sandy, loamy and clay. The work is based on measurement, simulation and model development. In the analysis, it is found that the average error percentages from the three developed models are 0.04%, 0.21% and 0.74% for sandy, loamy and clay soil, respectively. The effect of soil density with its electrical characteristics in terms of permittivity, propagation velocity and two-way wave travel time are also discussed. At the end of this paper, a soil density prediction tool is developed using the empirical models introduced consisting the density and attenuation for each soil sample at frequency range of 1.7 GHz to 2.6 GHz.