MAXimum Feasible Subsystem Recovery of Compressed ECG Signals

Electrocardiograph (ECG) signals are recorded continuously to monitor the health of potential cardiovascular disease (CVD) patients, leading to large amounts of data. An efficient way to acquire and compress signals would reduce bandwidth requirements for transmission and reduce memory and power requirements at the monitoring device. Compressive Sensing (CS) is an efficient method for ECG compression. However, the existing CS sparse recovery algorithms have small critical sparsity, which means that acceptable signal recovery requires many measurements. In this paper, two MAXimum Feasible Subsystem (MAX-FS)-based recovery algorithms that have shown good performance in speech compression are investigated for recovery of compressed ECG signals from the MIT-BIH Arrhythmia database. The two MAX-FS-based methods provide better recovery of compressed ECG signals than conventional recovery algorithms such as Smoothed '0 Norm (SL0) and Basis Pursuit (BP) with almost 47:5% and 30% reduction in the required number of measurements, respectively.