UC Irvine

In the increasingly digitized world, the intersection of data utilization and privacy protection presents significant challenges and opportunities. This dissertation explores the concept of Differential Privacy (DP), a framework that promises robust privacy protections while allowing the utility of data in diverse applications. Our research addresses the translation of DP from a theoretical construct into practical tools that can be integrated into real-world systems, focusing on DNS resolution and resource allocation.

One of the core advancements presented in this work is the development of a differentially private DNS resolution method that significantly reduces tracking accuracy rates with a provable guarantee. This is complemented by a prototype that the public can easily install on their local machines.

In the domain of resource allocation, we introduce novel differentially private mechanisms designed for environments such as cloud computing, virtual machine allocation, and network bandwidth management. These mechanisms not only ensure the confidentiality of sensitive metadata but also maintain system performance by integrating noise distribution techniques that optimize the trade-off between privacy protection and resource utility. This part of the study provides a comprehensive analysis of how differential privacy can be pragmatically applied to manage resources efficiently while adhering to stringent privacy standards, showcasing empirical results that support the feasibility of these approaches.

Additionally, the research broadens the scope of privacy-enhancing technologies beyond DP, exploring their application in machine learning.

Through rigorous empirical studies and innovative system design, this dissertation not only contributes to the academic field but also aims to influence real-world practices by enhancing the privacy and utility of systems in which large volumes of personal data are processed.The implications of this dissertation offer directions for future work in securing digital interactions and promoting a safer, more transparent digital environment. We anticipate the widespread adoption of privacy-preserving technologies across multiple sectors, promoting a balanced approach to data privacy that is adaptable to the changing digital landscape.