Enhancing Embedded IoT Systems for Intrusion Detection Using a Hybrid Model

Intrusion detection in Internet of Things (IoT) networks is a critical challenge due to the dynamic and evolving nature of cyber threats. To address this issue, we propose a novel hybrid intrusion detection algorithm that combines the deep autoencoder-based intrusion detection system (DAIDS) with recurrent neural networks (RNNs). The hybrid model aims to leverage the strengths of both approaches to enhance the accuracy and robustness of intrusion detection in IoT networks. The DAIDS component employs a deep autoencoder to capture complex spatial patterns in the network traffic data and calculate the reconstruction error as an anomaly score. The RNN component focuses on capturing temporal dependencies within sequential traffic data, providing a comprehensive understanding of network behavior over time. The output of the RNN is integrated into the DAIDS framework, augmenting the feature representation and enabling better anomaly detection. Experimental evaluation is conducted on a diverse IoT network traffic dataset, and the results showcase the effectiveness of the proposed hybrid DAIDS-RNN model. Comparative analysis demonstrates that the hybrid approach outperforms stand-alone DAIDS and RNN models in terms of accuracy, precision, recall, and F1-score. The hybrid model’s ability to identify both known and evolving attack patterns highlights its potential to enhance IoT intrusion detection systems. This study contributes to the advancement of intrusion detection techniques for IoT networks by introducing a hybrid approach that marries spatial and temporal insights for improved detection accuracy and adaptability to evolving threats. The proposed hybrid DAIDS-RNN model demonstrates its applicability in real-world IoT security scenarios, paving the way for more effective with 99.64 % accuracy and robust intrusion detection systems.