Baihui Jiang
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Baihui Jiang has a total of 4 co-authors. Here are the top co-authors by number of shared publications:
- Jia Liao
College of Big Data and Internet, Shenzhen Technology University, Shenzhen 518118, China
1 shared publication - Lei Ning
Shenzhen Technology University
1 shared publication - Liming Chen
Electric Power Research Institute of China Southern Power Grid, Guangzhou, China
1 shared publication - Peng Zhao
College of Big Data and Internet, Shenzhen Technology University, Shenzhen 518118, China
1 shared publication
Latest Publications
- Open Access
- 18 September 2023
- Published by MDPI AG in Electronics
- Vol. 12 (18) , 3938
Unmanned Aerial Vehicle-Assisted Federated Learning Method Based on a Trusted Execution Environment
In the face of increasing concerns around privacy and security in the use of unmanned aerial vehicles (UAVs) for mobile edge computing (MEC), this study proposes a novel approach to secure UAV-assisted federated learning. This research integrates a trusted execution environment (TEE) into UAV-assisted federated learning and proposes a robust aggregation algorithm based on cosine distance, denoted as CosAvg. This study further designs and evaluates a TEE-based federated learning model, comparing its resource overhead with other secure aggregation frameworks, like homomorphic encryption (HE) and differential privacy (DP). Experimental results indicate a significant reduction in resource overhead for TEE against DP and HE. Moreover, the proposed CosAvg algorithm demonstrated superior robustness against adversarial scenarios, maintaining high accuracy in the presence of malicious clients. The integration of TEE and the CosAvg algorithm provides a secure and robust solution for UAV-assisted federated learning, effectively defending both gradient inversion attacks and byzantine attacks.
- 18 September 2023
- Published by MDPI AG in Electronics
- Vol. 12 (18) , 3938
Open AccessUnmanned Aerial Vehicle-Assisted Federated Learning Method Based on a Trusted Execution Environment
In the face of increasing concerns around privacy and security in the use of unmanned aerial vehicles (UAVs) for mobile edge computing (MEC), this study proposes a novel approach to secure UAV-assisted federated learning. This research integrates a trusted execution environment (TEE) into UAV-assisted federated learning and proposes a robust aggregation algorithm based on cosine distance, denoted as CosAvg. This study further designs and evaluates a TEE-based federated learning model, comparing its resource overhead with other secure aggregation frameworks, like homomorphic encryption (HE) and differential privacy (DP). Experimental results indicate a significant reduction in resource overhead for TEE against DP and HE. Moreover, the proposed CosAvg algorithm demonstrated superior robustness against adversarial scenarios, maintaining high accuracy in the presence of malicious clients. The integration of TEE and the CosAvg algorithm provides a secure and robust solution for UAV-assisted federated learning, effectively defending both gradient inversion attacks and byzantine attacks.