Hybrid Distributed Deep-GAN Intrusion Detection System in IoT with Autoencoder

Internet of things integrates intelligent and smart devices in the surrounding environment to form dynamic heterogeneous networks. Hence, it is a time-consuming process for the IDS model to detect the anomaly behavior and a challenging task to provide security to the IoT networks. In this paper, the authors develop a hybrid distributed deep learning algorithm integrated with GAN (HDDGAN-IDS) and data mining techniques to detect intrusion attacks. In these proposed DDGAN-IDS, they deploy wrapper and filter-based flower pollination method in feature selection to reduce training time and avoids the overfitting of data and an auto encoder for feature extraction and dimensionality reduction which expedite the processing speed and finally the GAN network model performs classification. The experimental results prove that the HDDGAN-IDS algorithm provides better intrusion detection performance with respect to higher accuracy, precision, recall, f-measure, and lower false positive rate (FPR) compared to the existing deep learning algorithms.