Optimizing IoT Attack Detection in Edge AI: A Comparison of Lightweight Machine Learning Models and Feature Reduction Techniques

Kód výsledku v IS VaVaI

<a href="https://www.isvavai.cz/riv?ss=detail&h=RIV%2F00216305%3A26220%2F26%3A0198650" target="_blank" >RIV/00216305:26220/26:0198650 - isvavai.cz</a>

Výsledek na webu

<a href="https://link.springer.com/chapter/10.1007/978-3-032-00642-4_19" target="_blank" >https://link.springer.com/chapter/10.1007/978-3-032-00642-4_19</a>

DOI - Digital Object Identifier

<a href="http://dx.doi.org/10.1007/978-3-032-00642-4_19" target="_blank" >10.1007/978-3-032-00642-4_19</a>

Jazyk výsledku

angličtina

Název v původním jazyce

Optimizing IoT Attack Detection in Edge AI: A Comparison of Lightweight Machine Learning Models and Feature Reduction Techniques

Popis výsledku v původním jazyce

This paper investigates machine learning driven cyberattack detection in Internet of Things networks. It tackles challenges posed by high-dimensional data and devices with limited resources. The study focuses on feature reduction methods to improve Edge AI efficiency. It compares feature selection techniques, such as Random Forest importance and Recursive Feature Elimination, with feature extraction methods, including Principal Component Analysis and Linear Discriminant Analysis. Several lightweight models are evaluated: Decision Tree, Random Forest, Logistic Regression, Multi-Layer Perceptron, and LightGBM. These models are tested using the CICIoMT2024 dataset for both binary and multi-label classification tasks. Performance is measured by accuracy, precision, recall, F1-score, and inference time on a workstation and a Raspberry Pi. The results reveal that feature selection outperforms feature extraction with appropriate frameworks. Decision Tree and Random Forest achieve the best result: 99.89% accuracy in binary classification and 99.61% in multi-label tasks when using Random Forest feature selection with five selected features. On the Raspberry Pi, Decision Tree stands out with inference times of 11.94 s for binary tasks and 25.73 s for multi-label tasks, making it suitable for edge computing. This research provides a practical guide for enhancing Internet of Things security across resource-constrained devices.

Název v anglickém jazyce

Optimizing IoT Attack Detection in Edge AI: A Comparison of Lightweight Machine Learning Models and Feature Reduction Techniques

Popis výsledku anglicky

This paper investigates machine learning driven cyberattack detection in Internet of Things networks. It tackles challenges posed by high-dimensional data and devices with limited resources. The study focuses on feature reduction methods to improve Edge AI efficiency. It compares feature selection techniques, such as Random Forest importance and Recursive Feature Elimination, with feature extraction methods, including Principal Component Analysis and Linear Discriminant Analysis. Several lightweight models are evaluated: Decision Tree, Random Forest, Logistic Regression, Multi-Layer Perceptron, and LightGBM. These models are tested using the CICIoMT2024 dataset for both binary and multi-label classification tasks. Performance is measured by accuracy, precision, recall, F1-score, and inference time on a workstation and a Raspberry Pi. The results reveal that feature selection outperforms feature extraction with appropriate frameworks. Decision Tree and Random Forest achieve the best result: 99.89% accuracy in binary classification and 99.61% in multi-label tasks when using Random Forest feature selection with five selected features. On the Raspberry Pi, Decision Tree stands out with inference times of 11.94 s for binary tasks and 25.73 s for multi-label tasks, making it suitable for edge computing. This research provides a practical guide for enhancing Internet of Things security across resource-constrained devices.

Druh

D - Stať ve sborníku

CEP obor

—

OECD FORD obor

20202 - Communication engineering and systems

Projekt

<a href="/cs/project/VK01030019" target="_blank" >VK01030019: Interaktivní kontrolní seznamy pro efektivní testování kybernetické bezpečnosti</a><br>

Návaznosti

P - Projekt vyzkumu a vyvoje financovany z verejnych zdroju (s odkazem do CEP)<br>S - Specificky vyzkum na vysokych skolach

Rok uplatnění

2025

Kód důvěrnosti údajů

S - Úplné a pravdivé údaje o projektu nepodléhají ochraně podle zvláštních právních předpisů

Název statě ve sborníku

Availability, Reliability and Security. ARES 2025. Lecture Notes in Computer Science, vol 15998

ISBN

978-3-032-00642-4

ISSN

—

e-ISSN

—

Počet stran výsledku

18

Strana od-do

325-342

Název nakladatele

Springer, Cham

Místo vydání

Ghent, Belgium

Místo konání akce

Gent

Datum konání akce

11. 8. 2025

Typ akce podle státní příslušnosti

WRD - Celosvětová akce

Kód UT WoS článku

—