Integrating CNN-LSTM Networks with Statistical Filtering Techniques for Intelligent IoT Intrusion Detection

In the rapidly evolving landscape of the Internet of Things (IoT), securing networked devices against malicious intrusions is of paramount importance. This paper presents a novel approach for Intelligent IoT Intrusion Detection, leveraging integrated Convolutional Neural Network (CNN) and Long Short...

Full description

Saved in:
Bibliographic Details
Published in2024 8th International Conference on Smart Cities, Internet of Things and Applications (SCIoT) pp. 189 - 195
Main Authors Imani, Fatemeh, Kargar, Masoud, Assadzadeh, Alireza, Bayani, Ali
Format Conference Proceeding
LanguageEnglish
Published IEEE 14.05.2024
Subjects
Accuracy
Anomaly Detection
Deep Learning
Filtering
Internet of Things
Intrusion detection
IoT Intrusion Detection
Network Traffic Analysis
Noise reduction
Privacy
Robustness
Smart cities
Online AccessGet full text

Cover

More Information
Summary:In the rapidly evolving landscape of the Internet of Things (IoT), securing networked devices against malicious intrusions is of paramount importance. This paper presents a novel approach for Intelligent IoT Intrusion Detection, leveraging integrated Convolutional Neural Network (CNN) and Long Short-Term Memory (LSTM) networks, along with Statistical Filtering techniques for denoising. By synergistically combining deep learning architectures with statistical filtering, our methodology aims to enhance the accuracy and robustness of intrusion detection systems in IoT environments. Empirical evaluations on diverse datasets, including the Edge-IIoTset Dataset and CoAP-DoS Dataset, demonstrate the effectiveness of our approach, achieving competitive accuracy rates of 99.93% and 94.99% respectively. Furthermore, our model showcases promising computational efficiency, making it suitable for deployment on resource-constrained IoT devices. These results underscore the significance of our research in advancing the state-of-the-art in IoT privacy, paving the way for more resilient and adaptive intrusion detection systems.
DOI:10.1109/SCIoT62588.2024.10570107