Details

Autor(en) / Beteiligte

• Luong, Duc - Thuan
• Nguyen, Minh - Hoang
• Nguyen, Hong - Duc
• Tran Thi, Thanh - Thuy
• Nguyen, Nam - Hoang
• Hoang, Trong - Minh

Titel

A Study on the Efficiency of ML-Based IDS with Dimensional Reduction Methods for Industry IoT

Ist Teil von

• 2023 RIVF International Conference on Computing and Communication Technologies (RIVF), 2023, p.481-486

Ort / Verlag

IEEE

Erscheinungsjahr

2023

Quelle

IEEE Electronic Library (IEL)

Beschreibungen/Notizen

• The complexity and difficulty of detecting cyber-attacks due to the fast increase in linked devices and apps in the Internet of Things make the current security concerns more challenging. To deal with these problems, many attack detection models have been developed recently. Machine learning-based intrusion detection systems (IDSs) have become more popular due to their efficiency and scalability. Otherwise, the emergence of edge computing technologies has presented numerous novel opportunities for applications to achieve objectives such as min-imizing latency, accelerating data processing, and strengthening barriers against potential attacks from the IoT environment, in that deploying IDSs on edge computing can prevent attacks early and minimize damage to the network. However, machine learning-based IDS on edge devices face a significant challenge in concurrently achieving high accuracy and low complexity. We need a machine learning model suitable for reasonable sample parameters while ensuring accuracy. This study proposes and examines the common ML-based IDS models with dimensional reduction techniques to appropriate edge computing solutions. We use the Edge-IIoT dataset for evaluating the system's per-formance with various sample sizes. The experimental findings show that the accuracy can be up to 99% of multiple attack detection rates.