Internet of things security: A multi-agent-based defense system design

Abstract

The increasing Internet of Things (IoT) network complexity and sophisticated Distributed Denial of Service (DDoS) attacks at machine speeds make accurate and timely detection and mitigation of these attacks a challenging activity. This study presents a multi-agent-based system design (MAS-IoT) against DDoS attacks in the IoT network. The MAS-IoT consists of different types of agents which communicate using Advanced Encryption Standard (AES). In the context of the study, DDoS attacks were detected using a long-short-term memory (LSTM)-based model (LSTM-IoT) developed based on the CIC-IoT-2022 dataset with a 99.48% accuracy rate. The detection time of the LSTM-IoT and its time complexity were calculated and compared using similar methods mentioned in the literature. The results demonstrate the effectiveness of the LSTM-IoT in accurately detecting DDoS attacks. However, the ability to use it effectively and on time is vital to counter real-time attacks. The MAS-IoT system enables this with minimum human intervention.