DeepGuard: Enhancing Intrusion Detection Systems for Smart Agricultural IoT Networks Using Deep Learning

Abstract

With the rapid growth of smart agricultural networks, ensuring the security of these networks has become a crucial challenge. In particular, denying users the service using attacks poses a significant threat to the availability and reliability of such networks. This paper presents a security delivering system specifically designed for smart agricultural networks with the help of advanced learning-based neural methods. The proposed security delivering system leverages neural based networks that rely on a mathematical formulation that aids in modifying a given shape called convolution, to analyze and classify network data collected at the edge layer of the smart agriculture network. By training the CNN algorithm on a comprehensive dataset consisting of normal network traffic and various types of DDoS attacks, the system learns to accurately detect and predict DDoS attacks in real time. To assess the performance of the suggested system, numerous experiments were carried out using a representative dataset of smart agricultural IoT network traffic. The outcomes prove that the CNN-based IDS achieves high accuracy and low false positive rates in detecting DDoS attacks, effectively mitigating the effects of these strikes on the network. Comparisons made with accuracy: 0.9957 vs Test accuracy: 0.995667 indicates that the model can educate itself and predict incoming network samples, without memorizing the existing collection of L3 in network stack-based values delivered to the system.