Enhancing IoT Network Security through AIDriven Intrusion Detection with Hybrid AutoencoderGAN Fusion

The rapid growth of Internet of Things (IoT)IoT devices has changed how we connect and interact with our environment. This advancement has also complicated network security. As IoT networks grow, data integrity and confidentiality become more important. This study integrates a hybrid autoencoder and Generative Adversarial Network (GAN) model to use AI to solve security problems. The UNSW-NB15 dataset, a benchmark for network IDSs, is used to rigorously evaluate the proposed model. An autoencoder and GAN combination offers promising IDS for IoT networks. The autoencoder component excels at capturing concise and illustrative data representations, enabling it to spot network anomalies. The GAN component improves the model's ability to generate synthetic data samples that closely resemble network traffic patterns, making it easier to identify complex and novel attacks. The UNSW-NB15 dataset is known for its extensive network traffic data, making it ideal for testing the proposed artificial intelligence-based IDS. The proposed hybrid model achieves 99.13% accuracy. The result outperforms traditional DL methods. The proposed model's increased precision shows its ability to detect and address network anomaly risks.  The proposed system is evaluated using precision, recall, F1-Score, AUC-ROC, MCC, and Cohen's Kappa. The thorough evaluation shows that the Proposed hybrid autoencoder-GAN model provides a complete and efficient IoT network security solution in all dimensions. In conclusion, artificial intelligence-based intrusion detection and a hybrid model that combines autoencoder and generative adversarial network techniques improve IoT network security. The model's UNSW-NB15 dataset accuracy shows its ability to protect IoT ecosystems from many threats. This study helps improve the reliability and robustness of networks, boosting confidence in the growing world of interconnected devices.