Indian Journal of Science and Technology

Abstract

This paper presents the time variability analysis of photoplethysmogram biometric identification system. There have been few researches discussing about the effectiveness of PPG signal as a biometric identification system in different time instances. PPG signals of 5 subjects were obtained from MIMIC II Waveform Database, version 3, part 3 with a sampling frequency of 125 Hz. The signals were pre-process using low pass Butterworth filter. Then, discriminating features were extracted from the PPG waveform in varying time instances (different days). Finally, this PPG samples were classified using commonly known classification techniques for person identification. Based on experimentation results, PPG signals when using LMT and FT gives identification rates of 96% for both classifiers. For sensitivity and specificity test, both LMT and FT give the accuracy of 0.96 and 0.01. The precision test gives the result of 0.962 for LMT and 0.964 for FT. Thus, this outcome suggests the feasibility and robustness of PPG signals as a biometric modality in different time instances.
Keywords: Biometric, FT, LMT, PPG Signal, Person Identification, Time Variability