ECG signal processing for long-term healthcare monitoring in body sensor networks

This work focuses on body sensor networks (BSNs) based ECG signal processing for long-term healthcare monitoring. A body sensor network is a wireless sensor network consisting of various implantable or wearable biosensors and an external device as the network coordinator. Benefitting from the miniature-size and biocompatible wireless sensor nodes, body sensor networks can provide long-term, ubiquitous, and low-cost healthcare monitoring, with their interference to user's daily life reduced to the minimum.;Since most of the sensors are powered by batteries, energy efficiency is crucial for the lifetime and usability of a body sensor network. As Medium Access Control (MAC) is one of the most important factors that affect the energy efficiency of wireless communication, two energy-efficient MAC protocols, H-MAC and BSN-MAC are introduced in this work. H-MAC, a Time Division Multiple Access (TDMA) based medium access control protocol specially tailored for body sensor networks, aims to improve BSNs' energy efficiency by exploiting heartbeat rhythm to perform TDMA time synchronization. Using H-MAC, biosensors can achieve time synchronization without having to turn on their radios to receive periodic timing information from a central network coordinator, so that energy cost for time synchronization can be completely eliminated, and the network lifetime can be prolonged. The second MAC protocol proposed is BSN-MAC, which is an adaptive, feedback-based, and IEEE 802.15.4-compatible MAC protocol. It exploits the feedback information from the deployed sensors to form a closed-loop control of the MAC layer parameters. A control algorithm is proposed to enable the BSN coordinator to adjust parameters of the IEEE 802.15.4 superframe to achieve both high energy efficiency and low latency on energy critical nodes.;In the ECG signal processing section, we first propose a scheme that utilizes the body activity context information obtained from a body sensor network to help detect QRS complex, the most significant waveform in ECG signals, in a daily ambulatory environment. Body activity information is used to select the optimal leads and QRS complex detector, so that best QRS complex detection performance can be achieved under environments with different Signal Noise Ratios (SNRs). With the QRS complex located, a Hidden Markov Model (HMM) based technique is developed to perform further detailed ECG segmentation. In order to make HMMs adapt promptly to the temporal variations and reduce the misalignment errors, a body sensor network based active HMM parameter adaptation algorithm is presented. Instead of a single generic model, multiple individualized HMMs are used to improve the temporal adaptability. Once the ECG signals are segmented, clinical tests can be performed, such as Heart Rate Recovery (HRR) and ST segment depression analysis during Exercise Testing (ET). Then a Multivariate Autoregressive (MAR) based sensor fusion technique is introduced to improve the ECG processing reliability and accuracy by taking advantages of combining sensory data from heterogeneous biosensors in the network.;Since no satisfying multi-physiological parameter database is available to support the full spectrum study of body sensor networks, there is a need to build a customized hardware platform to collect biosignals. With the platform, various biosensors can communicate wirelessly and physiological parameters can be obtained from heterogeneous biosensors simultaneously and flexibly. A hardware platform designed for various biosignal acquisitions is discussed. The major components on the PCB and schematic are introduced.