picture:
Federated learning model based on blockchain
see more
Credits: Nan SUN, Wei WANG, Yongxin TONG, Kexin LIU
In the Internet of Things, network devices are more vulnerable to various intrusion attacks. Most existing algorithms are trained centrally, which may incur external communication costs and privacy leaks. Furthermore, traditional model training methods are unable to identify new types of unlabeled attacks.
To address these issues, a research team led by Wei WANG published their new search October 15, 2024 at Frontiers of Computing co-published by Higher Education Press and Springer Nature.
The team proposed a federated and distributed intrusion detection method, using the information contained in labeled data as prior knowledge to discover new types of unlabeled attacks. The detection method is verified and tested on the public dataset.
Compared with existing research results, the proposed method can ensure training security and discover new types of attacks.
In the research, a blockchain-based federated learning architecture is established. All participating entities perform model training locally and upload the model parameters to the blockchain. A collaborative model parameter verification mechanism and a proof-of-stake consensus mechanism are adopted, excluding malicious entities from the training process. The blockchain technique is introduced into the training architecture to ensure secure and distributed coordination of federated training.
To detect unknown attack types during local model training, the whole model training process includes three stages: pre-training phase, new attack discovery phase, and model training phase overall. An end-to-end clustering algorithm is used in each entity to distinguish different attack types, adopting the dissimilarity of spatio-temporal characteristics of the dataset. The experiments are performed in the AWID dataset. Experimental data show that compared with existing research methods, the proposed method can better guarantee training security and discover new types of intrusion attacks.
Future work could focus on developing a computationally efficient consensus mechanism capable of supporting the real-time requirements of IoT.
DOÏ: 10.1007/s11704-023-3026-8
Newspaper
Frontiers of Computing
Research method
Experimental study
Research subject
Not applicable
Article title
Federated blockchain-based learning for intrusion detection for the Internet of Things
Article publication date
15-October-2024
Disclaimer: AAAS and EurekAlert! are not responsible for the accuracy of press releases published on EurekAlert! by contributing institutions or for the use of any information via the EurekAlert system.
