Enhancing IoT Network Security by Anomaly Detection and Intrusion Prevention Using Gannet Optimization-Based Adaptive Deep Capsule Network

Document Type

Article

Publication Title

International Journal of Computational Intelligence Systems

Abstract

The grouping of various interconnected devices along distinct applications leads the Internet of Things (IoT) to more vulnerability to security threats that affect the security of the data. Thus, Intrusion Detection Systems (IDS) are needed for IoT to mitigate cyber threats. A proven performance is offered by the deep learning models in detecting network traffic, minimizing the effects of cyberattacks, and providing enhanced security to IoT devices. Thus, this paper aims the implementation of anomaly identification and network intrusion prevention in IoT systems. Three main steps are involved in the developed framework. The first step is to get the necessary data from the publicly accessible data source. Once the necessary data is collected, the best and most appropriate weighted features are obtained from the input data. The optimal weighted features are obtained with the aid of a newly introduced Improved Gannet Optimization Algorithm (IGOA), which is responsible for optimizing the weight necessary to fuse the features to form the weighted fused features to assist in the upcoming intrusion detection procedure. To find the anomaly in the network, the weighted fused features are given as input to run via the Adaptive Deep Capsule Network (ADCapsNet). The generated IGOA is used to tune the hyper-parameter in the ADCapsNet framework to increase the detection performance. Then, the necessary actions are taken to prevent these intrusions from the network. In the end, the implemented model is evaluated by contrasting it with various traditional anomaly detection models.

DOI

10.1007/s44196-025-00940-2

Publication Date

12-1-2025

This document is currently not available here.

Share

COinS