Metadata-Hiding Group Chat System Based on Cloud

With rising privacy concerns, hiding metadata, including timestamps, message frequency, and volume, has become essential in group chat systems based on cloud. Adversaries can exploit this data to infer user identities and social relationships. Existing techniques based on Distributed Point Functions (DPF) offer strong privacy but face two key limitations: (1) lack of flexibility in selecting multiple recipients, and (2) vulnerability to corruption through denial-of-service (DoS) attacks. In this paper, we propose a metadata-hiding group chat system based on cloud that addresses these limitations by enhancing the traditional DPF framework. Our construction enables an metadata-hiding group chat mechanism through a single operation that writes to multiple database channels simultaneously. To defend against DoS attacks, we design a flexible access control mechanism that incoporates secret sharing control techniques, specifically Private Access Control Lists (PACL), ensuring that only authorized writes are accepted. We implement a prototype of our system and demonstrate through experiments that it reduces key size by up to 93.04% and computational cost by 99.37% compared to tree-based DPF schemes. These improvements make our approach practical for real-world, lowlatency, and metadata-hiding group chat applications in the cloud.