Separation of Heartbeat Waveforms of Simultaneous Two-Subjects Using Independent Component Analysis and Empirical Mode Decomposition

Abstract

Vital sign monitoring using continuous wave (CW) microwave Doppler radar is gaining attention due to its simpler architecture and fewer signal processing chains. Existing literature focuses on utilizing the direction of arrival (DOA) technique of CW radar when the subjects are within the angular spacing limit. However, when two subjects cross the angular spacing limit for DOA estimation of the radar then the DOA technique becomes ineffective. To address this challenge, this research work focuses on testing the efficacy of two signal processing approaches [empirical mode decomposition (EMD) and independent component analysis with the joint approximation diagonalization of the eigenmatrices (ICA-JADE)] for the experimental scenarios when the subjects are within the beamwidth of the CW radar. After isolating the individual heartbeat waveforms using two different signal processing approaches it was compared with the Biopac ECG recorded heartbeat signal. Experimental results demonstrated that the ICA-JADE method superseded the performance of the EMD technique with an accuracy of 92.57% in all repeated measurements.