Search papers, labs, and topics across Lattice.
This paper evaluates seven pre-trained CNN architectures for IoT DDoS detection using the CICDDoS2019 dataset and image-based traffic representation, focusing on performance, reliability, latency, training cost, and interpretability. The study uses metrics like MCC, Youden Index, confidence intervals, Grad-CAM, and SHAP to assess the models. Results show that DenseNet and MobileNet architectures offer a good balance of detection performance, reliability, and class-consistent attribution, with DenseNet169 excelling in reliability and interpretability, and MobileNetV3 providing a latency-accuracy trade-off.
DenseNet169 not only nails IoT DDoS detection, but also offers reliability and interpretability that actually align, making it a standout choice.
Distributed denial-of-service (DDoS) attacks threaten the availability of Internet of Things (IoT) infrastructures, particularly under resource-constrained deployment conditions. Although transfer learning models have shown promising detection accuracy, their reliability, computational feasibility, and interpretability in operational environments remain insufficiently explored. This study presents an explainability-aware empirical evaluation of seven pre-trained convolutional neural network architectures for multi-class IoT DDoS detection using the CICDDoS2019 dataset and an image-based traffic representation. The analysis integrates performance metrics, reliability-oriented statistics (MCC, Youden Index, confidence intervals), latency and training cost assessment, and interpretability evaluation using Grad-CAM and SHAP. Results indicate that DenseNet and MobileNet-based architectures achieve strong detection performance while demonstrating superior reliability and compact, class-consistent attribution patterns. DenseNet169 offers the strongest reliability and interpretability alignment, whereas MobileNetV3 provides an effective latency-accuracy trade-off for fog-level deployment. The findings emphasize the importance of combining performance, reliability, and explainability criteria when selecting deep learning models for IoT DDoS detection.
