Decentralized CNN Integrated Federated Learning for Intrusion Detection System

As the technological landscape evolves rapidly in Industry 5.0, cyberspace is expanding, creating more diverse and complex data. This development not only makes industries more sustainable, resilient, and human-centric but also results in a larger attack surface for potential cybercrime. Researchers proposed Machine Learning(ML)-based Intrusion Detection Systems(IDS) to tackle the scenarios. As most of the data born are decentralized, the traditional centralized ML-based IDS are unable to cope with the present scenarios. In this paper, We have proposed a privacy-preserving federated learning model that collaborates with Convolution Neural Network(CNN) to improve intrusion detection outcomes. Here, collaboration and model training are performed without violation of data security and privacy. The benchmark dataset UNSW-NB15 is used to conduct the experiment and measure the performance of the model. The experimental result achieves better performance to address the existing problems.