A hybrid CNN+LSTM-based intrusion detection system for industrial IoT networks

Engineering Science and Technology an International Journal, Journal Year: 2023, Volume and Issue: 38, P. 101322 - 101322

Published: Jan. 6, 2023

The Internet of Things (IoT) ecosystem has proliferated based on the use internet and cloud-based technologies in industrial area. IoT technology used industry become a large-scale network increasing amount data number devices. Industrial (IIoT) networks are intrinsically unprotected against cyber threats intrusions. It is, therefore, significant to develop Intrusion Detection Systems (IDS) order ensure security IIoT networks. Three different models were proposed detect intrusions by using deep learning architectures Convolutional Neural Network (CNN), Long Short Term Memory (LSTM), CNN + LSTM generated from hybrid combination these. In study conducted UNSW-NB15 X-IIoTID datasets, normal abnormal determined compared with other studies literature following binary multi-class classification. model attained highest accuracy value for intrusion detection both datasets among models. architecture an 93.21% classification 92.9% dataset while same 99.84% 99.80% dataset. addition, accurate success implemented regarding types attacks within was evaluated.

Language: Английский

---

A review of ensemble learning and data augmentation models for class imbalanced problems: Combination, implementation and evaluation

Expert Systems with Applications, Journal Year: 2023, Volume and Issue: 244, P. 122778 - 122778

Published: Dec. 10, 2023

Class imbalance (CI) in classification problems arises when the number of observations belonging to one class is lower than other. Ensemble learning combines multiple models obtain a robust model and has been prominently used with data augmentation methods address problems. In last decade, strategies have added enhance ensemble methods, along new such as generative adversarial networks (GANs). A combination these applied many studies, evaluation different combinations would enable better understanding guidance for application domains. this paper, we present computational study evaluate prominent benchmark CI We general framework that evaluates 9 Our objective identify most effective improving performance on imbalanced datasets. The results indicate can significantly improve find traditional synthetic minority oversampling technique (SMOTE) random (ROS) are not only selected problems, but also computationally less expensive GANs. vital development novel handling

Language: Английский

---

Inference of Wildfire Causes From Their Physical, Biological, Social and Management Attributes

Earth s Future, Journal Year: 2025, Volume and Issue: 13(1)

Published: Jan. 1, 2025

Abstract Effective wildfire prevention includes actions to deliberately target different causes. However, the cause of an increasing number wildfires is unknown, hindering targeted efforts. We developed a machine learning model ignition across western United States on basis physical, biological, social, and management attributes associated with wildfires. Trained from 1992 2020 12 known causes, overall accuracy our exceeded 70% when applied out‐of‐sample test data. Our more accurately separated ignited by natural versus human causes (93% accuracy), discriminated among 11 classes human‐ignited 55% accuracy. attributed greatest percentage 150,247 for which source was unknown equipment vehicle use (21%), lightning (20%), arson incendiarism (18%).

Language: Английский

---

DIDS: A Deep Neural Network based real-time Intrusion detection system for IoT

Decision Analytics Journal, Journal Year: 2022, Volume and Issue: 5, P. 100142 - 100142

Published: Nov. 17, 2022

Language: Английский

---

TNN-IDS: Transformer neural network-based intrusion detection system for MQTT-enabled IoT Networks

Computer Networks, Journal Year: 2023, Volume and Issue: 237, P. 110072 - 110072

Published: Oct. 17, 2023

The Internet of Things (IoT) is a global network that connects large number smart devices. MQTT de facto standard, lightweight, and reliable protocol for machine-to-machine communication, widely adopted in IoT networks. Various devices within these networks are employed to handle sensitive information. However, the scale openness make them highly vulnerable security breaches attacks, such as eavesdropping, weak authentication, malicious payloads. Hence, there need advanced machine learning (ML) deep (DL)-based intrusion detection systems (IDS). Existing ML-based IoT-IDSs face several limitations effectively detecting activities, mainly due imbalanced training data. To address this, this study introduces transformer neural network-based system (TNN-IDS) specifically designed MQTT-enabled proposed approach aims enhance activities TNN-IDS leverages parallel processing capability Transformer Neural Network, which accelerates process results improved attacks. evaluate performance system, it was compared with various IDSs based on ML DL approaches. experimental demonstrate outperforms other terms activity. achieved optimum accuracies reaching 99.9% activities.

Language: Английский

---

Toward Enhanced Attack Detection and Explanation in Intrusion Detection System-Based IoT Environment Data

IEEE Access, Journal Year: 2023, Volume and Issue: 11, P. 131661 - 131676

Published: Jan. 1, 2023

Securing the Internet of Things (IoT) against cyber threats is a formidable challenge, and Intrusion Detection Systems (IDS) play critical role in this effort. However, lack transparent explanations for IDS decisions remains significant concern. In response, we introduce novel approach that leverages blending model attack classification integrates counterfactual Local Interpretable Model-Agnostic Explanations (LIME) techniques to enhance explanations. To assess effectiveness our approach, conducted experiments using recently introduced CICIoT2023 IoTID20 datasets. These datasets are real-time large-scale benchmark IoT environment attacks, offering realistic challenging scenario captures intricacies intrusion detection dynamic environments. Our experimental results demonstrate improvements accuracy compared conventional methods. Furthermore, proposed provides clear interpretable insights into factors influencing decisions, empowering users make informed security choices. Integrating explanation enhances reliability systems. Therefore, work represents advancement detection, robust defense cyber-attacks data.

Language: Английский

---

GA-mADAM-IIoT: A New Lightweight Threats Detection in the Industrial IoT Via Genetic Algorithm with Attention Mechanism and LSTM on Multivariate Time Series Sensor Data

Sensors International, Journal Year: 2024, Volume and Issue: 6, P. 100297 - 100297

Published: Sept. 4, 2024

Language: Английский

---

Class overlap handling methods in imbalanced domain: A comprehensive survey

Multimedia Tools and Applications, Journal Year: 2024, Volume and Issue: 83(23), P. 63243 - 63290

Published: Jan. 11, 2024

Language: Английский

---

GPT and Interpolation-Based Data Augmentation for Multiclass Intrusion Detection in IIoT

IEEE Access, Journal Year: 2024, Volume and Issue: 12, P. 17945 - 17965

Published: Jan. 1, 2024

The absence of essential security protocols in Industrial Internet Things (IIoT) networks introduces cybersecurity vulnerabilities and turns them into potential targets for various attack types. Although machine learning has been used intrusion detection the IIoT, datasets with representative data common attacks IIoT network traffic are limited often imbalanced. Data augmentation techniques address these problems by creating artificial classes fewer samples. In this work, we evaluate use when training models based on data. We compare Generative Pre-trained Transformers (GPT) variations Synthetic Minority Over-sampling TEchnique (SMOTE) their capability to enhance performance. examine performance five algorithms trained augmented original non-augmented dataset. To ensure a fair comparison, evaluated algorithms' different scenarios using same test dataset, which does not contain synthetic results show need systematic evaluation before employing augmentation, as its impact classification depends algorithm, data, technique. While deep neural benefit from eXtreme Gradient Boosting (XGBoost), achieved superior between all classifiers (with F1-Score over 91%), didn't have any improvement generated GPT-based methods shows such (especially GReaT) generate invalid both numerical categorical features way that leads degradation multiclass classification.

Language: Английский

---

Efficient Training: Federated Learning Cost Analysis

Big Data Research, Journal Year: 2025, Volume and Issue: unknown, P. 100510 - 100510

Published: Feb. 1, 2025

Language: Английский

---