Dynamic network security leveraging efficient CoviNet with granger causality-inspired graph neural networks for data compression in cloud IoT Devices

The increasing adoption of Internet of Things (IoT) devices has led to the centralization of data on cloud servers, offering enhanced memory capacity and advanced administrative capabilities. However, this change introduces significant security risks. Therefore, a novel method is proposed to predict and secure dynamic networks in cloud-based IoT systems using a Granger Causality Inspired Graph Neural Network integrated with Efficient coviNet (GCIGNN-ENet). The proposed GCIGNN-ENet framework classifies network traffic into Benign, Bot, Brute Force, DDoS, DoS, Heartbleed, Infiltration, Port Scan, and Web Attack. EfficientcoviNet replaces the causality-inspired layer in GCIGNN, and incorporating dense layers from ENetV2 for improved performance. Despite its capabilities, GCIGNN-ENet struggles to dynamically optimize security thresholds for cloud IoT systems. To address this, the Local Neighbour Spider Monkey Optimization (LNSMO) algorithm is employed to enhance the accuracy of classifiers and adaptability in predicting network security threats. The proposed DNSEGCN-CIOTD framework is implemented in Python and evaluated using performance metrics, like Accuracy, Precision, Recall, F-score, and Computational Time. The simulation results demonstrate that the DNSEGCN-CIOTD achieves 23.12 %, 21.23 %, and 21.32 % higher accuracy, 23.12 %, 21.23 % and 21.32 % higher recall and 23.54 %, 22.18 % and 23.65 % higher F-score when compared to the existing methods, such as Transfer Learning Method for IDS on Cloud IoT Devices utilizing Optimized CNN (TIDS-CIOT-CNN), Ranking Security of IoT-dependent Smart Home Consumer Devices (RS-IOT-SHCD), Rotating Behind Security: An Enhanced Authentication Protocol for IoT-enabled Devices in Distributed Cloud Computing Architecture (RSAPIOTD-DCA).