A Novel Network Intrusion Detection Based on Variational Long Short-Term Memory and Federated Learning Under Cloud Computing Environment

This paper proposes a novel network intrusion detection method based on Variational Long Short-Term Memory (VLSTM) and Federated Learning (FL), aiming to effectively enhance the performance of network intrusion detection in cloud computing environments. In the cloud computing environment, network traffic data has the characteristics of being massive, complex and distributedly stored. Traditional intrusion detection methods are challenged by the issues of insufficient data privacy protection and model generalization ability. To address these problems, we first utilized the VLSTM network to extract features and model the network traffic data. VLSTM, by introducing a variational mechanism, can better capture the time series features and potential uncertainty information in the network traffic data, thereby improving the ability to identify intrusion behaviors. Then, combined with the federated learning framework, we conducted collaborative training on the data distributed across different cloud nodes, without requiring the data to be centralized in a single central node. This effectively solved the problem of data privacy protection and fully utilized the advantages of distributed data, enhancing the model’s generalization ability and detection accuracy. Experimental results show that this method outperforms the existing mainstream intrusion detection methods in key indicators such as detection rate and false alarm rate, providing a new and effective solution for network intrusion detection in the cloud computing environment, and has significant theoretical and practical value.