Hierarchical Classification of Botnet Using Lightweight CNN

Applied Sciences, Journal Year: 2024, Volume and Issue: 14(10), P. 3966 - 3966

Published: May 7, 2024

This paper addresses the persistent threat of botnet attacks on IoT devices, emphasizing their continued existence despite various conventional and deep learning methodologies developed for intrusion detection. Utilizing Bot-IoT dataset, we propose a hierarchical CNN (HCNN) approach featuring three levels classification. The HCNN approach, presented in this paper, consists two networks: non-hierarchical network. network works by combining features obtained at higher level with those its descender. combined information is subsequently fed into following to extract descendant nodes. overall 1790 parameters, introducing an additional 942 parameters existing backbone. classification comprise binary normal vs attack first level, followed 5 classes second 11 third level. To assess effectiveness our proposed evaluate performance metrics such as Precision (P), Recall (R), F1 Score (F1), Accuracy (Acc). Rigorous experiments are conducted compare both models state-of-the-art approaches, providing valuable insights efficiency addressing devices.

Language: Английский

---

Synthetic Data Generation and Impact Analysis of Machine Learning Models for Enhanced Credit Card Fraud Detection

IFIP advances in information and communication technology, Journal Year: 2024, Volume and Issue: unknown, P. 362 - 374

Published: Jan. 1, 2024

Language: Английский

---

TSSAN: Time-Space Separable Attention Network for Intrusion Detection

IEEE Access, Journal Year: 2024, Volume and Issue: 12, P. 98734 - 98749

Published: Jan. 1, 2024

With the continuous evolution of novel network attacks, traditional Intrusion Detection Systems (IDSs) have commonly employed Deep Neural Networks (DNNs) for intrusion detection. However, effectiveness a DNN in this respect is closely related to quality training data set, and large-scale traffic are difficult label accurately. Therefore, some challenges still need be addressed detect attacks. In paper, we introduce Time-Space Separable Attention Network (TSSAN) TSSAN utilizes depth wise separable convolution time-space self-attention mechanism effectively extract temporal spatial features. By extracting common features from unlabeled data, significantly enhanced detection performance rare attack types. Experimental evaluations were conducted using UNSW-NB15 CICIDS-2017 datasets. Meticulous experiments evaluating individual components model rigorously carried out dataset. unsupervised learning experiment, our method achieved 0.86 0.92 f1score two semi-supervised learning, experiment showed that performed better than deep when labelled gradually reduced.

Language: Английский

---

An Entanglement-Aware Middleware for Digital Twins

ACM Transactions on Internet of Things, Journal Year: 2024, Volume and Issue: 5(4), P. 1 - 25

Published: Oct. 14, 2024

The development of the Digital Twin (DT) approach is tilting research from initial approaches that aim at promoting early adoption to sophisticated attempts develop, deploy, and maintain applications based on DTs. In this context, we propose a highly dynamic distributed ecosystem where containerized DTs co-evolve with an orchestration middleware. provide digitalized representations targeted physical systems, while middleware monitors re-configures deployed in light application constraints, available resources, quality cyber-physical entanglement. First, lay out reference scenario. Then, discuss limitations current identify set requirements shape both Subsequently, describe blueprint architecture meets those requirements. Finally, report empirical evidence feasibility effectiveness proof-of-concept implementation proposed ecosystem.

Language: Английский

---