Zero Trust Unified Access Control based on Multi-Scenario Fusion Training

The basic idea of federated learning is to aggregate the local model parameters of all participants to obtain a global model, which is more general and has better performance. However, the existing federated learning suffers from the problem of a unique model, where all participants train their local models based on the same global model, which is not suitable for multi-scenario training. In this paper, we propose a new decentralized federated learning model, which moves the aggregation work from the central server to each participant's local device, enabling joint training of the model without sharing the datasets across different scenarios. Moreover, the model parameters are aggregated based on model similarity, which enables end-to-end feature learning for multi-scenario fusion and thus achieves unified access control for zero-trust on end devices. This paper applies the proposed model to the identification of abnormal terminal traffic and tests the recognition performance of the model using the NSL-KDD dataset. The results show that the proposed model achieves a five-classification recognition accuracy of up to 92.76
^%
on the NSL-KDD dataset, demonstrating good recognition performance.