Securing Federated Learning Enabled NWDAF Architecture With Partial Homomorphic Encryption
Authors:
- 12 July 2023
- journal article
- research article
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Networking Letters
- Vol. 5 (4) , 299-303
Abstract
Network data analytics function (NWDAF), introduced to provision data analytics and machine learning model training in the 5G core network, is expected to be an essential functional entity and play a significant role in the emerging AI-native 6G wireless network. However, refining the NWDAF architecture to support machine learning (ML) model sharing among multiple NWDAFs with distributed data sources and different privacy constraints remains a major challenge. To address this challenge, we propose a federated learning enabled NWDAF architecture with Partial Homomorphic Encryption to secure ML model sharing with privacy preserving. Simulation results demonstrate the feasibility of our proposed architecture.Keywords
This publication has 8 references indexed in Scilit:
- A Decentralized Collaborative Learning Approach in 5G+ Core NetworksIEEE Network, 2023
- Homomorphic Encryption-Based Privacy-Preserving Federated Learning in IoT-Enabled Healthcare SystemIEEE Transactions on Network Science and Engineering, 2022
- Differential Privacy for Deep and Federated Learning: A SurveyIEEE Access, 2022
- The Road Towards 6G: A Comprehensive SurveyIEEE Open Journal of the Communications Society, 2021
- Deep Leakage from Gradients2020
- Deep Learning with Differential Privacy2016
- Model Inversion Attacks that Exploit Confidence Information and Basic Countermeasures2015
- The MNIST Database of Handwritten Digit Images for Machine Learning Research [Best of the Web]IEEE Signal Processing Magazine, 2012