Electronic Theses and Dissertations

Date

2024

Document Type

Dissertation

Degree Name

Doctor of Philosophy

Department

Computer Science

Committee Chair

Kan Yang

Committee Member

Christos Papadopoulos

Committee Member

Myounggyu Won

Committee Member

Xiaofei Zhang

Abstract

Managing digital credentials—such as certificates, identification documents, health records, cryptographic keys, and intellectual property (e.g., neural network models or trade secrets)—presents substantial challenges in both \textit{credential issuance} and \textit{credential presentation}, particularly around centralization, privacy, and scalability. Centralized credential issuance through systems like Certificate Authorities (CAs) creates a single point of failure, thereby undermining trust and elevating security vulnerabilities. Similarly, credential presentation via centralized identity services like Single Sign-On (SSO) entrusts the control of user credentials to third parties, exacerbating privacy risks. While user-centric models, such as self-sovereign identity (SSI), offer individuals greater control over their data, these systems grapple with scalability and interoperability issues, especially as resource-constrained devices often lack the computational capacity to execute zero-knowledge proofs (ZKPs), which are crucial for privacy-preserving credential presentation and verification. Furthermore, managing computational-based credentials—which involve proving the correctness of computations tied to intellectual property (IP), such as machine learning models—presents even greater complexities. Verifying the integrity and ownership of these computational assets requires sophisticated cryptographic proofs, such as ZKPs, to ensure both security and privacy. These proofs must verify the results of complex computations (e.g., training a neural network) without exposing the training data or model, further complicating the credential management process. Furthermore, managing credentials that require complex cryptographic proofs—such as verifying the integrity and ownership of intellectual property, like neural network models—introduces additional complexities in ensuring both security and privacy in the credential management process. This dissertation proposes a decentralized trust infrastructure using blockchain, threshold cryptography, and zero-knowledge proofs to create a secure, scalable, and privacy-preserving credential management system. It decentralizes credential issuance and presentation to avoid single points of failure, enhance privacy, and achieve scalable, verifiable computation for trust and interoperability. Key solutions include PKChain for decentralized issuance via threshold cryptography, SilentProof for privacy-preserving presentation with ZKPs, and zkAdHoc for verifying complex credentials, like neural network models, without exposing sensitive data. This framework addresses centralization, privacy, scalability, and complex computation, advancing privacy, scalability, and trust across digital environments.

Comments

Data is provided by the student.

Library Comment

Dissertation or thesis originally submitted to ProQuest.

Notes

Embargoed until 11-08-2026

Download

Available for download on Sunday, November 08, 2026

Share

COinS