电子科技2026，Vol.39Issue(2)：96-104,9.DOI:10.16180/j.cnki.issn1007-7820.2026.02.012
基于RSA同态加密的联邦学习隐私保护方案
A Privacy Protection Scheme for Federated Learning Based on RSA Homomorphic Encryption
摘要
Abstract
In view of the problems of the current federated learning security aggregation schemes,such as gradi-ent information leakage,large computational overhead of the participants,and lack of integrity protection for the mod-el,this study proposes a federated learning privacy protection scheme based on RSA(Rivest-Shamir-Adleman)ho-momorphic encryption.The problem of gradient data leakage is solved by constructing an efficient and secure RSA homomorphic encryption algorithm,and the decryption calculation is implemented through the central server to reduce the computational overhead of the participants.Based on the immutability and evidence preservation protection fea-tures of blockchain,the key data during the data training process is stored on the chain for evidence preservation,a-chieving the full life cycle maintenance of model data.Through security analysis,it can be known that the proposed scheme has the ability to resist collusion attacks and the indistinguishability of ciphertexts.Through the analysis and comparison in terms of performance and efficiency,it can be known that the proposed scheme has certain advantages over the traditional scheme.关键词
联邦学习/隐私保护/同态加密/区块链/RSA/安全聚合/存证保护/密文不可去分性Key words
federated learning/privacy protection/homomorphic encryption/blockchain/RSA/secure aggrega-tion/certificate protection/indivisibility of ciphertext分类
信息技术与安全科学引用本文复制引用
GUAN Guilin,CAI Huimin,ZHI Ting,CAO Yang,DING Hongxin,HUANG Jiang,DAI Yang..基于RSA同态加密的联邦学习隐私保护方案[J].电子科技,2026,39(2):96-104,9.基金项目
国家重点研发计划(2023YFC3806001) （2023YFC3806001）
海南省重大科技计划(ZDKJ2021051) （ZDKJ2021051）
贵州省科技支撑计划(2023MA6DN7B8X22057)National Key R&D Program of China(2023YFC3806001) （2023MA6DN7B8X22057）
Hainan Province Major Science and Technology Plan(ZDKJ2021051) （ZDKJ2021051）
Guizhou Province Science and Technology Support Program(2023MA6DN7B8X22057) （2023MA6DN7B8X22057）
