Date of Award
Doctor of Philosophy (PhD)
Smart health monitoring uses real-time monitored data to support diagnosis, treatment, and health decision-making in modern smart healthcare systems and benefit our daily life. The accurate health monitoring and prompt transmission of health data are facilitated by the ever-evolving on-body sensors, wireless communication technologies, and wireless sensing techniques. Although the users have witnessed the convenience of smart health monitoring, severe privacy and security concerns on the valuable and sensitive collected data come along with the merit. The data collection, transmission, and analysis are vulnerable to various attacks, e.g., eavesdropping, due to the open nature of wireless media, the resource constraints of sensing devices, and the lack of security protocols. These deficiencies not only make conventional cryptographic methods not applicable in smart health monitoring but also put many obstacles in the path of designing privacy protection mechanisms.
In this dissertation, we design dedicated schemes to achieve secure data collection and analysis in smart health monitoring. The first two works propose two robust and secure authentication schemes based on Electrocardiogram (ECG), which outperform traditional user identity authentication schemes in health monitoring, to restrict the access to collected data to legitimate users. To improve the practicality of ECG-based authentication, we address the nonuniformity and sensitivity of ECG signals, as well as the noise contamination issue. The next work investigates an extended authentication goal, denoted as "wearable-user pair" authentication. It simultaneously authenticates the user identity and device identity to provide further protection. We exploit the uniqueness of the interference between different wireless protocols, which is common in health monitoring due to devices' varying sensing and transmission demands, and design a "wearable-user pair" authentication scheme based on the interference. However, the harm of this interference is also outstanding. Thus, in the fourth work, we use wireless human activity recognition in health monitoring as an example and analyze how this interference may jeopardize it. We identify a new attack that can produce false recognition result and discuss potential countermeasures against this attack. In the end, we move to a broader scenario and protect the statistics of distributed data reported in mobile crowd sensing, a common practice used in public health monitoring for data collection. We deploy differential privacy to enable the indistinguishability of workers' locations and sensing data without the help of a trusted entity while meeting the accuracy demands of crowd sensing tasks.
Huang, Pei, "Secure Data Collection and Analysis in Smart Health Monitoring" (2021). All Dissertations. 2910.