The Study of PPG and APG Signals for Biometric Recognition

Abstract

Nowadays, the numbers of identity theft victims are increasing year by year and caused financial losses. This issue needs to be dealt with before it becomes worst. Many ways has been done in order to decrease the number of identity theft victims. For examples, password is needed in order for the valid user to access the applications or services and identifications card are used for entry in premises. However, these approaches have limitations such as easy to be forgotten and lost. Recently, bio-signals are getting attention among researchers since it marks our vital parts of the body and used in the unimodal biometric system. Therefore, this study proposes of a more secure mechanism by using PPG and APG signal for biometric recognition system. PPG signals data will be collected from 10 different subjects by using an Easy Pulse sensor data acquisition device. Then, in order to obtain APG signals, the process of signal transformation was conducted. Next, preprocessing was applied to remove the unwanted signal or noises. After that, the features of the PPG and APG signals were extracted. Finally, these PPG and APG samples undergo the classification process by using classifiers to identify individuals. Based on the experimentation results, PPG signal obtained 84% identification rates as compared to the result 94% of APG signal when using Bayes Network. The next classifier used is Multilayer Perceptron (MLP) which gives result of 84% and 92% of PPG signal and APG signal respectively. The other two classifiers used are Sequential Minimal Optimization (SMO) and K-Nearest Neighbors (IBk). The achieved result for PPG signal is 90% while for APG signal the result is 96% when using SMO classifier. Lastly, the obtained result for IBk classifier is 92% for PPG signal and 94% for APG signal. The outcome of this project proved that multimodal biometric can be performed by using PPG and APG signal since everyone has different PPG and APG signal.