Enhancing IoT Security: A Deep Learning Approach with Feedforward Neural Network for Detecting Cyber Attacks in IoT

Abstract

A new era of connectedness has been ushered in by the increasing number of Internet of Things (IoT) devices, which present both enormous security issues and limitless opportunities for creativity. With the use of a deep learning-powered intrusion detection system (IDS), this research aims to improve IoT security. An extensive dataset of different cyberattack kinds was used to train and test a Feedforward Neural Network (FNN) for its ability to detect intrusions using the CIC-IoT2023 dataset. The FNN achieved excellent accuracy, an F1 score, and a precision score, which are encouraging results. This shows the system's capability to differentiate between legitimate and fraudulent network traffic and illustrates its potential value in protecting IoT ecosystems. However, there are certain restrictions, such as the necessity for continuing optimization and the representativeness of the dataset. This research provides knowledge regarding the efficiency of deep learning-based IDS, which is an essential step toward strengthening IoT security. This work lays the groundwork for continued initiatives to guarantee the reliability and safety of linked IoT devices in a constantly shifting threat environment as the IoT environment develops.