Files in this item



application/pdfLi_Qingkun.pdf (4MB)
(no description provided)PDF


Title:An energy-efficient hardware system for robust and reliable heart rate monitoring
Author(s):Li, Qingkun
Advisor(s):Iyer, Ravishankar K.; Kalbarczyk, Zbigniew T.
Department / Program:Electrical & Computer Eng
Discipline:Electrical & Computer Engr
Degree Granting Institution:University of Illinois at Urbana-Champaign
Subject(s):reconfigurable hardware
biomedical monitoring
fault tolerance
embedded hardware
heart rate monitoring
Abstract:Cardiac arrhythmia, one of the most common causes of death in the world today, is not always effectively detected by regular examinations, as it usually occurs infrequently and suddenly. Therefore, real-time, continuous monitoring of the heart rate is needed to detect arrhythmia problems sooner and prevent their severe consequences. To make continuous monitoring possible and give it widespread acceptance, a portable heart rate monitoring system must have three key characteristics: (1) accuracy, (2) portability, and (3) long battery life. Previous studies have focused on addressing these problems separately, either improving the accuracy of the monitoring algorithm or the efficiency of the underlying hardware. This thesis proposes a robust and reliable heart rate monitoring system (RRHMS), in which both algorithm accuracy and hardware efficiency are considered. As a result, algorithmic optimizations are exploited to enable further hardware efficiency. In the RRHMS, robust heart rate monitoring is achieved by extracting heart rates from both electrocardiogram (ECG) and arterial blood pressure (ABP) signals and fusing them based on the signal qualities. Therefore, accurate heart rate data can be provided continuously, even when one signal is severely corrupted. Algorithmic optimizations are applied to merge the separate ECG and ABP processing steps into shared ones, which allows shared hardware modules and hence low-area (portable) hardware design. Also, an embedded hardware architecture framework is proposed for the design of the RRHMS hardware system. Coarse-grained functional units (FUs) can be easily added or removed in this framework, allowing for application-specific hardware optimization. Further, the application invariant properties are used to achieve low-overhead fault tolerance in the FUs to enhance reliability. Both ASIC and FPGA implementations of the RRHMS are able to accurately detect heart rates in real time while consuming only 1/2870 and 1/923 of the energy required by the Android implementation.
Issue Date:2015-03-16
Rights Information:Copyright 2015 Qingkun Li
Date Available in IDEALS:2015-07-22
Date Deposited:May 2015

This item appears in the following Collection(s)

Item Statistics