Year: 2025
Type: article
Abstract: Threshold multi-party fully homomorphic encryption (TMFHE) schemes enable efficient computation to be performed on sensitive data while maintaining privacy. These schemes allow a subset of parties to perform threshold decryption of evaluation results via a distributed protocol without the need for a trusted dealer, and provide a degree of fault tolerance against a set of corrupted parties. However, existing TMFHE schemes can only provide correctness and security against honest-but-curious parties. We construct a compact TMFHE scheme based on the Learning with Errors (LWE) problem. The scheme applies Shamir secret sharing and share resharing to support an arbitrary t-out-of-N threshold access structure, and enables non-interactive reconstruction of secret key shares using additive shares derived from the current set of online participants. Furthermore, the scheme implements commitment and non-interactive zero-knowledge (NIZK) proof techniques to verify the TMFHE operations. Finally, our experiments demonstrate that the proposed scheme achieves active security against malicious adversaries. It overcomes the limitation of existing TMFHE schemes that can only guarantee correct computation under passive semi-honest adversaries.
Source: Applied Sciences
DOI: 10.3390/app15094745
Authors: Yuqi Xie, Ruwei Huang, Junbin Qiu
Institution: Guangxi University
Language: English
Subfield: Artificial Intelligence
Field: Computer Science
Domain: Physical Sciences
Open Access status: gold
Estimated APC paid: $2,896
OpenAlex ID: W4409764239