Network security framework for IoT data transmission using disentangled cascaded graph convolutional networks with Eurasian lynx optimizer and advanced encryption

Abstract

A robust network security framework ensures secure data transmission in Internet of Things (IoT) networks, protecting devices and sensitive information from cyber threats. However, the diversity and scale of IoT devices pose significant challenges in maintaining consistent, efficient, and real-time security across all endpoints. The paper proposes a novel DC-GCN – ELO architecture for secure IoT intrusion detection using network traffic data from the UNSW-NB15 and NSL-KDD datasets. After that, preprocessing of this information is carried out based on Fuzzy Min-Max Neural Networks (FMMNN). It includes noise removal, dealing with uncertain data, as well as removing redundancy and the absence of records. The next step involves extraction of significant attributes through the use of Multidimensional Graph Transformer Networks (MGTN). DC-GCN identifies attack patterns, ELO improves efficiency, and PE-DS maintains secure data transmission of IoT data. The model gives 99.40% and 99.20% accuracy scores on UNSW-NB15 and NSL-KDD datasets, respectively, along with high precision, recall, and F1 scores. The use of a framework with high data integrity, low latency, and reduced communication overhead results in effective and reliable intrusion detection and secure IoT data transmission.