Append-only Authenticated Data Sets based on RSA accumulators for transparent log system

Abstract

Authenticated Data Structures (ADS) play a vital role in ensuring data authenticity in cloud computing environments, particularly in Public Key Infrastructure (PKI). Despite their importance, significant concerns remain about cloud computing security, such as the risk of hijacking certification authorities to issue fraudulent certificates or tampering with log entries. While Certificate Transparency (CT) improves server certificate credibility, CT log servers remain vulnerable, risking data integrity and trust. This paper proposes the Append-only Authenticated Data Sets (AADS) model, which enhances cloud computing security by supporting append-only properties and fork consistency in transparent logs, preventing malicious modifications and preserving divergent log views. Formally defined using RSA accumulators, the model includes efficient algorithms for initialization, data addition, search, and evidence verification, ensuring security under strong RSA and adaptive root assumptions while reducing space complexity. Additionally, Append-only Authenticated Dictionaries (AAD) offer better space and communication efficiency. A prototype transparent log system demonstrates how these structures mitigate identity authentication risks caused by missing append-only features and fork consistency. Key challenges remain, including scalability, cryptographic weaknesses, and coordinated attacks by compromised log servers and malicious entities.