Low-Energy ECG Processing for Accurate Features' Extraction in Wireless Body Sensor Networks

Abstract

Wireless body sensor networks (WBSNs) represent an attractive low-cost infrastructure for e-health applications. In depth, these wireless tiny devices can be deployed to gather, to process and transmit bio-signals providing remote real-time monitoring of patient health signs. Among applications that can be implemented using these sensors, the ECG signal processing for features' extraction to detect arrhythmia disease is attracting the focus of several recent research activities. In this context, the detection of Atrial Fibrillation episodes in ECG signal requires accurate analysis of RR irregularity and efficient detection of the absence of P-waves. However, the weak processing bandwidth and the limited energy in these sensors strongly limit their adequacy to be deployed for such application. This paper focuses on the design of a low-energy ECG processing scheme, for WBSN implementation, that accurately extracts the relevant ECG features required for atrial fibrillation detection and notify a remote server. The paper presents the performance analysis of this approach and shows the efficiency of the proposed scheme for energy aware ECG processing and its adequacy to be implemented in WBSN.