Attack detection in IoT-Fog computing environment using hybrid optimization-based Generative Adversarial Network

Abstract

Internet of Things (IoT) is considered the rapidly growing paradigm in the evolution of computing. A number of attack detection methods is introduced for detecting the attacks in IoT-Fog computing, but achieving significant result poses a complex task due to the requirements of the IoT devices, like distribution, low latency, limitation of resources, and scalability. Hence, an effective attack detection method is designed using the proposed Sail Fish Cat Spider-based Generative Adversarial Network for detecting the attacks in the IoT framework. The proposed SFCS is developed by the integration of Sail Fish Optimization (SFO) with Cat Swarm Optimization (CSO) and Spider Monkey Optimization (SMO), respectively. Here, the features are selected from the data using Minkowski distance, and the process of attack detection is accomplished using GAN. However, the deep learning classifier named GAN is trained using the proposed SFCS optimization algorithm. The deep learning classifier is more effective in detecting cyber-attacks in the IoT environment. However, the average values of the proposed SFCS-based GAN obtained by considering the metrics, recall, precision, and F-measure is specified with the values of 85.787%, 85.165%, and 85.247% for DoS attack, and recall, precision, and F-measure of 85.340%, 85.653%, and 85.243% for Fuzzers attack.