Deep Learning Approaches for DDoS Attack Detection in IoT: Knowledge Gaps and Experimental Analysis

The fast growth of Internet of Things (IoT) devices has substantially escalated the exposure of IoT networks to Distributed Denial-of-Service (DDoS) attacks, putting the availability and reliability of critical services at risk. This paper presents a thorough study of deep learning-based DDoS attack detection approaches for IoT environments. First, the existing literature is reviewed to identify the major research gaps of high-dimensional network traffic, manual hyperparameter selection, computational complexity, and limited model generalization. The Long Short-Term Memory (LSTM) network is then introduced due to its ability to capture long-term temporal dependencies in sequential network traffic. The performance of six deep learning models has been experimentally evaluated using the IoTID20 benchmark dataset. To ensure a fair comparison, the hyperparameters of each model are directly taken from the corresponding reference paper. The experimental results show that the LSTM model achieves the best performance, with an accuracy of 95.02%, higher than the other deep learning methods. The comparative analysis shows the benefits and constraints of the existing approaches and provides useful insights for future research work on intelligent IoT intrusion detection systems.