ABSTRACT
Edge computing can move storage and computing tasks from cloud data centers to the edge of networks, reducing latency. However, this carries security and privacy risks. Federated learning can protect privacy by transferring training models from edge compute nodes to local devices, which then train models on local datasets. But the parameters transmitted between local devices and edge nodes may contain raw data, which can be stolen. To address this, this paper proposes a FedAVG-based privacy-preserving federated learning (FPPFL) scheme to provide low latency and privacy protection. It optimizes key generation parameters with an improved Paillier homomorphic encryption algorithm, allowing for low-complexity exponential operations in the encryption stage, shortening data transmission time and protecting model parameters, while maintaining accuracy. Experiments demonstrate the improved Paillier algorithm achieves the same security level as benchmark schemes and outperforms them in computational complexity.
- K. Wang , ”Task Offloading With Multi-Tier Computing Resources in Next Generation Wireless Networks,” in IEEE Journal on Selected Areas in Communications, vol. 41, no. 2, pp. 306-319, Feb. 2023.Google ScholarCross Ref
- EmuraKeita, KatsumataShuichi, and WatanabeYohei, “Identity-based encryption with security against the KGC,” 2022.Google Scholar
- Z. H. Mahmood and M. K. Ibrahem, ”New Fully Homomorphic Encryption Scheme Based on Multistage Partial Homomorphic Encryption Applied in Cloud Computing,” 2018 1st Annual International Conference on Information and Sciences (AiCIS), Fallujah, Iraq, 2018, pp. 182-186.Google ScholarCross Ref
- R. Podschwadt, D. Takabi, and P. Hu, “SoK: Privacy-preserving Deep Learning with Homomorphic Encryption,” 2021.Google Scholar
- Alessandro Giuseppi, Lucrezia Della Torre, Danilo Menegatti, Francesco Delli Priscoli, Antonio Pietrabissa, and Cecilia Poli, ”An Adaptive Model Averaging Procedure for Federated Learning (AdaFed),” Journal of Advances in Information Technology, Vol. 13, No. 6, pp. 539-548, December 2022.Google ScholarCross Ref
- X. X. Tian, C. F. Sha, X. L. Wang, and A. Y. Zhou, “Privacy Preserving Query Processing on Secret Share Based Data Storage,” 2011.Google ScholarCross Ref
- Mart´ı , “Deep Learning with Differential Privacy,” ACM, 2016.Google Scholar
- Y. Wang, X. Liang, X. Hei, W. Ji, and L. Zhu, “Deep Learning Data Privacy Protection Based on Homomorphic Encryption in AIoT,” Mobile Information Systems, vol. 2021, no. 2, pp. 1–11, 2021.Google Scholar
- Z. Zhou, Y. Tian, and C. Peng, “Privacy-Preserving Federated Learning Framework with General Aggregation and Multiparty Entity Matching,” Wireless Communications and Mobile Computing, vol. 2021, pp. 1–14, 2021.Google Scholar
- Rukhin , “Statistical test suite for random and pseudorandom number generators for cryptographic applications, NIST special publication,” 2010.Google Scholar
Index Terms
- FPPFL: FedAVG-based Privacy-Preserving Federated Learning
Recommendations
Ubiquitous intelligent federated learning privacy-preserving scheme under edge computing
AbstractWith the rapid development of artificial intelligence (AI), combining machine learning (ML) and edge computing powered by big data has become a growing trend. However, under edge computing, ML faces problems such as limited energy ...
Highlights- Propose a architecture of federated learning with cloud–edge-device collaboration under edge computing to solve the problem of insufficient computing power ...
The security of a strong proxy signature scheme with proxy signer privacy protection
In 1996, Mambo et al. first introduced the concept of a proxy signature scheme, and discussed the delegation of the signing capability to a proxy signer. In 2001, Lee et al. constructed a strong non-designated proxy signature scheme. In 2002, Shum and ...
Learnable Privacy-Preserving Anonymization for Pedestrian Images
MM '22: Proceedings of the 30th ACM International Conference on MultimediaThis paper studies a novel privacy-preserving anonymization problem for pedestrian images, which preserves personal identity information (PII) for authorized models and prevents PII from being recognized by third parties. Conventional anonymization ...
Comments