ECG compression using wavelet-based compressed sensing with prior support information

Electrocardiogram (ECG) signal compression is a vital signal processing area, especially with the growing usage of wireless body sensor networks (WBSN). ECG signals need to be compressed for efficient storage or transmission. Traditional compression methods acquire the ECG signal and perform compression in the sensors, where most of the computations are performed. Therefore, sensors have significant complexity and power consumption. On the other hand, compressed sensing transfers the complexity from the encoder to the decoder, thus allowing for cheap, low-power sensors, with longer lifetime. In this paper, we propose the adaptive reduced-set matching pursuit with partially known support (ARMP-PKS) compressed sensing reconstruction algorithm. ARMP-PKS performs sparse signal reconstruction at very high accuracy and speed. Furthermore, it takes advantage of prior support information, which further improves the performance. Our proposed algorithm achieves 35 dB reconstruction SNR at 50% compression and about 22.5 dB at 25%, significantly faster than related algorithms.

Automated summary

This study proposes a compressed sensing reconstruction algorithm for ECG signals which transfers complexity from the encoder to the decoder, achieving high accuracy and speed in sparse signal reconstruction. By utilizing prior support information, the algorithm further improves performance, achieving a 35 dB reconstruction SNR at 50% compression and about 22.5 dB at 25%, significantly faster than related algorithms.