Augmented whale feature selection for IoT attacks: Structure, analysis and applications

Abstract

Smart connected appliances expand the boundaries of the conventional Internet into the new Internet of Things (IoT). IoT started to hold a significant role in our life, and in several fields as in transportation, industry, smart homes, and cities. However, one of the critical issues is how to protect IoT environments and prevent intrusions. Attacks detection systems aim to identify malicious patterns and threats that cannot be detected by traditional security countermeasures. In literature, feature selection or dimensionality reduction has been profoundly studied and applied to the design of intrusion detection systems. In this paper, we present a novel wrapper feature selection approach based on augmented Whale Optimization Algorithm (WOA), which adopted in the context of IoT attacks detection and handles the high dimensionality of the problem. In our approach, we introduce the use of both V-shaped and S-shaped transfer functions into WOA and compare the superior variant with other well-known evolutionary optimizers. The experiments are conducted using NBaIoT dataset; wherein, five datasets were sampled from the original dataset. The dataset represents real IoT traffic, which is drawn from the UCI repository. The experimental results show that WOA based on V-shaped transfer function combined with elitist tournament binarization method is superior over S-shaped transfer function and outperforms other well regarded evolutionary optimizers based on the obtained average accuracy, fitness, number of features, running time and convergence curves. Hence, we can conclude that the proposed approach can be deployed in IoT intrusion detection systems