Fog-enabled Intelligent Network Intrusion Detection Framework for Internet of Things Applications
Sujit Bebortta, Saneev Kumar Das, Sujata Chakravarty · 2023
With the continual growth in Internet of Things (IoT), several concerns regarding the security and privacy of the connected IoT devices arise. Attacks like Distributed Denial of Service (DDoS) and probing attacks cause specific damage to the underlying network infrastructure. In this paper, we propose an Equilibrium Optimization-based Artificial Neural Network (EO-ANN) Framework for detection of network attacks pertaining to IoT systems. The proposed approach leverages the dynamic mass balance technique to obtain the optimal number of features for training the ANN algorithm. To evaluate the effectiveness of the proposed EO-ANN model, its performance is compared with several baseline prediction models. Experimental findings demonstrate that the suggested approach has precision and recall values of 0.9389 and 0.9248, respectively, and an F1-score of 0.9371, which results in an overall accuracy of 93.4731%. In addition to these performance requirements, the model was implemented over the Fog computing platform and the corresponding metrics viz., delay and energy consumption were obtained towards execution of different workloads. Hence, the experimentations offered in this study show that the suggested EO-ANN model is effective and that it can be successfully validated for classifying network intrusion in densely deployed IoT networks.