Feature attention assisted convolutional stacked sparse auto-encoder model for intrusion detection in network function virtualization environment

Network function virtualization (NFV) in 5 G networks has recently received much attention. However, it generates numerous challenges while providing security in emerging technologies such as information, education, biotechnology, etc. NFV exploration has concentrated on intrusion detection because detecting an intrusion is necessary due to the wastage of resources and security threats. Therefore, an intrusion detection system called Feature Attention assisted Convolutional Stacked Sparse Auto-encoder (FA_CS²ANet) Model for Intrusion Detection in the NFV Environment has been proposed. To detect intrusions in the NFV network, the input data is first collected from a publicly available dataset, and then pre-processing is performed to remove the unwanted data using min-max normalization, standardization and missing value replacement. Next, feature selection is done to reduce the dimensionality issues using Chaotic Osprey Optimization (COO). After selecting the necessary features, the intrusions in NFVs are identified by using the deep learning-based FA_CS²ANet model, which is a combination of the Convolutional Neural Network (CNN) and Stacked Sparse Auto-encoder (SSAE) model. The simulation is completed using Python programming, and the results demonstrate that the suggested method outperforms existing methods with an accuracy of 93.12%. The intrusions are discovered, and the suggested method’s performance metrics for accuracy, precision, recall, and F-score are assessed.