A lightweight optimized intrusion detection system using machine learning for edge-based IIoT security

The Industrial Internet of Things (IIoT) attributes to intelligent sensors and actuators for better manufacturing and industrial operations. At the same time, IIoT devices must be secured from the potentially catastrophic effects of eventual attacks, and this necessitates real-time prediction and pr...

Full description

Saved in:
Bibliographic Details
Published in:Telecommunication systems 2024-11, Vol.87 (3), p.605-624
Main Authors: Tiwari, Ravi Shekhar, Lakshmi, D., Das, Tapan Kumar, Tripathy, Asis Kumar, Li, Kuan-Ching
Format: Article
Language:English
Subjects:
Accuracy
Actuators
Artificial Intelligence
Business and Management
Computer Communication Networks
Cybersecurity
Discriminant analysis
Industrial applications
Industrial Internet of Things
Internet of Things
Intrusion detection systems
IT in Business
Machine learning
Network latency
Particle swarm optimization
Payloads
Principal components analysis
Probability Theory and Stochastic Processes
Real time
Regression models
Telecommunications systems
Citations: Items that this one cites
Online Access:Get full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:The Industrial Internet of Things (IIoT) attributes to intelligent sensors and actuators for better manufacturing and industrial operations. At the same time, IIoT devices must be secured from the potentially catastrophic effects of eventual attacks, and this necessitates real-time prediction and preventive strategies for cyber-attack vectors. Due to this, the objective of this investigation is to obtain a high-accuracy intrusion detection technique with a minimum payload. As the experimental process, we have utilized the IIoT network security dataset, namely WUSTL-IIOT-2021. The feature selection technique Particle Swarm Optimization (PSO) and feature reduction techniques such as Principal Component Analysis (PCA), Linear Discriminant Analysis (LDA), and t-distributed stochastic neighbor embedding (t-SNE) are applied. Additionally, the Generalized Additive Model (GAM) and Multivariate Adaptive Regression Splines (MARS) are used to detect payloads that can interfere with the normal operation of an application. Both PSO and PCA combined with MARS have produced predictive results with an exceptional accuracy of 100%. Yet, the trained Machine Learning (ML) model is quantized with 4-bit and 8-bit, and it is deployed on Azure IoT Edge to simulate edge devices. Experimental results show that the latency of the model was reduced by 25% on quantization.
ISSN:1018-4864
1572-9451
DOI:10.1007/s11235-024-01200-y