Robust reconstruction for CS-based fetal beats detection

Due to its possible low-power implementation, Compressed Sensing (CS) is an attractive tool for physiological signal acquisition in emerging scenarios like Wireless Body Sensor Networks (WBSN) and telemonitoring applications. In this work we consider the continuous monitoring and analysis of the fetal ECG signal (fECG). We propose a modification of the low-complexity CS reconstruction SL0 algorithm, improving its robustness in the presence of noisy original signals and possibly ill-conditioned sensing/reconstruction procedures. We show that, while maintaining the same computational cost of the original algorithm, the proposed modification significantly improves the reconstruction quality, both for synthetic and real-world ECG signals. We also show that the proposed algorithm allows robust heart beat classification when sparse matrices, implementable with very low computational complexity, are used for compressed sensing of the ECG signal.