| dc.description.abstract | Embedded systems are ubiquitous, and more systems are continuing to be deployed through the recent development of the Internet of Things. Embedded systems are specialized computer systems that are designed to perform specific tasks, typically with a dedicated function or limited application domain. They are engineered to be small, efficient, and low-powered, and can be found in a wide range of devices and applications, including cars, medical equipment, home appliances, and industrial machines. Embedded systems play a critical role in many of the devices and systems that we rely on every day, and their importance is only expected to grow as the world becomes more connected and reliant on technology.
Security has often been a secondary consideration in the design and implementation of embedded systems, with a focus on functionality and performance taking priority. However, this has changed in recent years as the risks associated with insecure embedded systems have become more apparent. One of the main challenges is that many embedded systems have limited resources, such as memory and processing power, which can make it difficult to implement strong security measures. Additionally, these systems may have a long lifespan and may not receive security updates, which can leave them vulnerable to attacks. A newer concern is that many embedded systems are designed to be connected to the internet, the Internet of Things, or other networks, which can in-crease their attack surface. Attackers can exploit vulnerabilities in these systems to gain access to sensitive data or disrupt their operation. Hence, securing these systems is becoming increasingly important. The aim of this dissertation is to increase the security of software in embedded systems.
We propose the use of existing hardware primitives commonly found on modern CPUs in embedded systems, specifically hardware performance counters and watchdog timers to enhance the security of embedded system software. The first method is to train a model that can detect anomalous execution of software. The second method we propose in this work is a way to have the embedded software continuously authenticate to ensure proper execution. The final piece of the work is to enhance a traditional watchdog timer with hardware performance counters to make it more robust against malicious modifications. | |
