Smart IoT Attack Detection Through AI-Optimized Routing Methods

Abstract

The Internet of Things (IoT) model possesses much flexibility and mobility, hence the IoT system becomes more prone to security threats such as Distributed Denial-of-Service (DDoS) attacks due to its decentralized control, dynamic node moving, energy scarces and bandwidth limited. This development has demonstrated the capabilities of AI in providing improvements to performance of IoT network in terms of achieving faster rate, provide more throughput and achieve higher packet delivery ratio. Utilization of AI-based analytics technology with the use of adaptive methods can improve metrics such as End to End delay (E2E) and Average Received Packets (ARP), by improving response time and increasing intelligence in Intrusion Detection Systems in IoT based setups. In this paper, we use Feedforward Neural Networks (FFNN) and Convolutional Neural Networks (CNN) to detect malicious activities which can strengthen the capabilities of IDS in IoT routing. The suggested AI-optimized routing model outperforms existing models in detection accuracies, which are improved up to 82% and 85% with respective processing time of 18s and 17s. These findings demonstrate the potential to significantly improve security of IoT systems along with increasing overall network robustness and efficiency using this framework.