Self-adaptive reconstruction for compressed sensing based ECG acquisition in wireless body area network

Abstract

The compressed sensing (CS) has been demonstrated as a promising solution for low-cost signal acquisition in wireless body area network. In this paper, a novel signal reconstruction scheme based on adaptive dictionary and matched filtering in CS domain is proposed for the ECG acquisition. The proposed method selects adaptive overcomplete dictionary based on the QRS estimation of the compressed measurements in each frame. If a QRS complex is estimated in this frame, an adaptive overcomplete dictionary matching the QRS characteristics of this frame is selected for reconstruction, otherwise, a dictionary trained by segments without QRS complex is selected. The ECG frames whose estimated QRS complexes locate in several consecutive locations, the so-called region width, are considered as one category, and will be reconstructed by one overcomplete dictionary which is trained by similar ECG waves. Extensive experiments have been conducted, and the results well demonstrate the effectiveness for signal reconstruction as well as its advantages over some state-of-the-art algorithms.