Enhancing Detection Capacity and Security for Network-based Intrusion Detection Systems in Internet of Things

As the number of Internet of Things devices continues to grow, so does the risk of attacks, which significantly impacts financial aspects and the potential loss of sensitive data. This makes Network-based Intrusion Detection Systems (NIDS) using machine learning algorithms more critical than ever. However, a key concern is the input data for the machine learning models. Most models require large datasets to achieve high accuracy, but this data often needs to maintain privacy if it includes personal information. Additionally, the trustworthiness of devices in the system must be ensured, as unreliable devices can reduce the model's effectiveness or lead to significant data loss. To address these challenges, we propose using Generative Adversarial Networks (GAN) to enrich training data and combining Blockchain with Federated Learning (FL) to ensure data privacy and component validation. This approach results in a robust, efficient model that balances lightweight design with high reliability.