Electrical and Computer Engineering ETDs

Publication Date

8-30-2011

Abstract

Electrocardiogram (ECG) monitoring systems have evolved to the point where they are now portable and can monitor the patient 24/7 and transmit alerts and ECG data to parents and doctors as soon as a heart irregularity is detected. With the advances in these systems, there is a need for the incorporation of ECG coding systems to reduce the bandwidth used when data is transmitted and to incorporate methods to provide data recovery in the event of a transmission error. However, while ECG encoding systems for hospital or home care settings has been thoroughly researched, the application of ECG encoding systems to portable ECG monitoring systems where there is a much higher likelihood of noise interference during transmission of the data has not been fully investigated. The goal of this work is to develop a real-time ECG encoding system that requires low hardware and power usage, provides lossless signal compression, and provides recovery of as much data as possible in the event of data corruption of packets during transmission. An entropy based compression algorithm is developed based on the Huffman code which is then transformed to reversible variable length codes. This allows the data packets to be both frontward and backwards decodable allowing for greater data recovery in the event that portions of a packet are corrupted. The implementation is designed to be able to encode any sized bit width by utilizing a combination of 4, 6, or 8- bit entropy coders. Two separate encoding systems are investigated using the before-mentioned encoding algorithm. The first system recomputes the Reversible Variable Length Code (RVLC) tables periodically while the signal is being encoded in an effort to adapt to any changes in the signal. The second system uses a pre-calculated RVLC table that minimizes the delay and also significantly reduces the required hardware resources. We provide optimal, reconfigurable implementations for both systems. Furthermore, the effectiveness and error-resilient performance of both systems are validated on 12-bit and 16-bit ECG signals. The performance of the system is shown to be diagnostically lossless in noisy communications channels with significant bit errors. This represents a significant improvement over existing systems that do not employ the proposed error resilient encoding methods.

Keywords

Electrocardiography--Data processing., Error-correcting codes (Information theory)

Document Type

Thesis

Language

English

Degree Name

Electrical Engineering

Level of Degree

Masters

Department Name

Electrical and Computer Engineering

First Committee Member (Chair)

Pollard, Howard

Second Committee Member

Zarkesh-Ha, Payman

Share

COinS