Date of Award
Doctor of Philosophy (PhD)
The Power Delivery Network (PDN) is an essential universal sub-system for any electronic system to operate correctly. Modern electronic systems are growing exponentially in their functionalities and complexity, and so does the sophistication of the PDNs they must depend on. Consequently, these complex PDNs create many pathways for unintended interactions and expose the system to numerous power side-channel attacks. Recent works have shown that many such vulnerabilities can be exploited remotely, making them especially potent security threats to modern electronic devices with ubiquitous connectivity. However, facing emerging PDN-based attacks, both attackers and defenders tend to treat PDN as a black box and overlook it in the security analysis due to its analog/mixed-signal nature and lack of functional abstraction. On the other hand, previous investigations on PDN mainly focus on the trade-off between performance, efficiency, and supply noise. As a result, such a gap in security analysis for PDN exacerbates PDN becoming a server security concern in electronic systems. In this dissertation, we aim to address the above gap and answer the research question: What is the role of PDN in power side-channel attacks, and what are the gains and losses brought by PDN from a security perspective? We proposed systemic security analysis for PDN and achieved unique contributions to the field. To understand the role of PDN in power side-channel attacks, we propose a security-oriented PDN modeling framework. Based on the insights on fundamental mechanisms causing PDN-based vulnerabilities, I then explore the utility and risk brought by modern PDN. We reveal that the PDN impedance can be used to detect PCB anomalies. However, at the chip level, the PDN impedance-based signatures can also invalidate cloud computing security measures. In addition, as a risk, PDN can also create a new vulnerability by allowing attackers to attack touchscreens through charger noises.The broader scope of this dissertation focuses on the security analysis of hidden analog domain properties in the digital systems. IC designers have long enjoyed the benefits of simplifying circuit behaviors into corresponding digital abstraction. Although beneficial for productivity, confining the security analysis only to the digital domain is insufficient to protect against vulnerabilities manifesting as analog behaviors. The analog behaviors are due to either the analog modules (e.g., the PDN) or the increasing parasitic effects (e.g., A2 Trojan) in advanced nodes. The contributions of this dissertation on PDN analysis and detecting analog vulnerabilities provide unique insights into the field when facing more advanced nodes and complex electronic systems.
Shantanu Chakrabartty, Yier Jin, Neal Patwari, Ning Zhang,
Available for download on Wednesday, May 15, 2024