|Full text PDF:||http://arks.princeton.edu/ark:/88435/dsp016969z0853|
The proliferation and wide usage of embedded and mobile systems have greatly impacted our lives but have also brought new challenges in secure system design. With increasing functional complexity, software programmability, and network connectivity, the security of modern embedded and mobile computing systems shares many commonalities with general-purpose computing systems. However, their unique usage models as well as extreme size and battery-power constraints make it infeasible to directly borrow conventional security solutions from general-purpose computing systems. The task of securing embedded and mobile systems includes protecting the confidentiality, integrity, and availability of their network communication, data storage, and software content. This thesis focuses on three prominent categories of security threats facing embedded and mobile computing systems: network-related threats, malware and software vulnerabilities, and side-channel attacks. It discusses design techniques that can help secure the wireless communication, prevent and detect software exploits, and defend against side-channel attacks. These techniques are well suited for embedded and mobile computing systems in that they consume very low energy. This thesis tackles a special type of embedded system network personal healthcare systems. Such systems are composed of implantable and wearable medical devices and commonly used for diagnosing, monitoring, and treating various medical conditions. Through a comprehensive vulnerability assessment, we demonstrate the necessity and importance of enhancing the trustworthiness of these systems. We discuss and analyze suitable safety measures in response to various types of threats. This thesis also presents a medical security monitor that performs anomaly detection to capture potentially malicious wireless transmissions in personal healthcare systems. Upon detection of a malicious transmission, the monitor can jam the packets before they are acknowledged by the medical device, and thus protect it from malicious commands or contaminated data. A key benefit of this methodology is that it is applicable to existing medical devices with no hardware or software modifications required. Consequently, it leads to zero power overheads on these devices.