Abstract

A group of wirelessly communicating sensors that are placed inside, on or around a human body constitute a Wireless Body Area Network (WBAN). Continuous monitoring of vital signs through WBANs have a potential to revolutionize current health care services by reducing the cost, improving accessibility, and facilitating medical diagnosis. However, sensitive nature of personal health data requires WBANs to integrate appropriate security methods and practices. As limited hardware resources make conventional security measures inadequate in a WBAN context, this work is focused on alternative techniques based on Wireless Physical Layer Security (WPLS). More specifically, we introduce a symbiosis of WPLS and Compressed Sensing to achieve security at the time of sampling. We successfully show how the proposed framework can be applied to electrocardiography data saving significant computational and memory resources. In the scenario when a WBAN Access Point can make use of diversity methods in the form of Switch-and-Stay Combining, we demonstrate that output Signal-to-Noise Ratio (SNR) and WPLS key extraction rate are optimized at different switching thresholds. Thus, the highest key rate may result in significant loss of output SNR. In addition, we also show that the past WBAN off-body channel models are insufficient when the user exhibits dynamic behavior. We propose a novel Rician based off-body channel model that can naturally reflect body motion by randomizing Rician factor K and considering small and large scale fading to be related. Another part of our investigation provides implications of user's dynamic behavior on shared secret generation. In particular, we reveal that body shadowing causes negative correlation of the channel exposing legitimate participants to a security threat. This threat is analyzed from a qualitative and quantitative perspective of a practical secret key extraction algorithm.

Library of Congress Subject Headings

Body area networks (Electronics)--Security measures; Data encryption (Computer science); Medical records--Data processing--Security measures

Publication Date

4-24-2019

Document Type

Dissertation

Student Type

Graduate

Degree Name

Computing and Information Sciences (Ph.D.)

Advisor

Gill R Tsouri

Advisor/Committee Member

Daniel Phillips

Advisor/Committee Member

Sohail Dianat

Comments

This dissertation has been embargoed. The full-text will be available on or around 5/9/2020.

Campus

RIT – Main Campus

Plan Codes

COMPIS-PHD

Available for download on Friday, May 08, 2020

Share

COinS