Keywords

computer architecture, architecture security, secure processor, non-volatile memory, emerging memory technology

Abstract

Ensuring information security and scalable performance are critical in modern processor architecture designs. High-performance, data-intensive applications demand robust microarchitectures that are both secure-by-design and efficient. This dissertation explores new information leakage threats in modern processors and proposes secure-by-design architectures for emerging systems.

First, we examine vulnerabilities in existing microarchitectures, focusing on speculative execution's security implications. We reveal that branch predictor designs in modern processors are speculatively updated, leading to a novel speculative execution side channel. We propose a secure branch predictor architecture to mitigate this issue while maintaining performance benefits.

Second, we address challenges with emerging non-volatile memory (NVM) technologies. We identify performance-limiting characteristics of resistive-memory-based NVMs, particularly variable write latency. We present a processor-side framework to optimize NVM write latency and design a secure architecture compatible with the NVMe protocol to protect data in high-speed NVMe storage devices.

Lastly, we emphasize the importance of composability in security when analyzing modern microarchitectures. We explore side channels in secure processor architectures that provide hardware-based data confidentiality and integrity protection. Our research highlights the need to synergistically understand microarchitecture security as new mechanisms are integrated.

Overall, this dissertation enhances our understanding of security and performance in modern computing systems and offers solutions to their challenges. We highlight the importance of a holistic approach to security and performance to effectively integrate new technologies into modern systems.

Completion Date

2024

Semester

Summer

Committee Chair

Yao, Fan

Degree

Doctor of Philosophy (Ph.D.)

College

College of Engineering and Computer Science

Department

Electrical and Computer Engineering

Degree Program

Computer Engineering

Format

application/pdf

Identifier

DP0028893

Language

English

Rights

In copyright

Release Date

2-15-2025

Length of Campus-only Access

None

Access Status

Doctoral Dissertation (Open Access)

Campus Location

Orlando (Main) Campus

Accessibility Status

Meets minimum standards for ETDs/HUTs

Download

Share

COinS