Approach for Detecting Attacks on IoT Networks Based on Ensemble Feature Selection and Deep Learning Models.

Area of Science:

• Cybersecurity
• Machine Learning
• Network Security

Background:

• The proliferation of Internet of Things (IoT) devices has led to increased cybersecurity vulnerabilities.
• IoT attacks pose significant threats to both organizations and individuals, necessitating advanced detection methods.

Purpose of the Study:

• To propose and evaluate an effective approach for detecting attacks in IoT networks.
• To assess the performance of deep learning models enhanced by ensemble feature selection.

Main Methods:

• Utilized ensemble feature selection, combining filter methods (variance threshold, mutual information, Chi-square, ANOVA, L1) with Recursive Feature Elimination (RFE).
• Evaluated the impact of selected features on various Deep Learning (DL) models, including CNN, RNN, GRU, and LSTM.
• Tested the proposed method on the IoT-Botnet 2020 dataset, measuring detection accuracy, precision, recall, F1-measure, and False Positive Rate (FPR).

Main Results:

• All evaluated DL models demonstrated high performance, achieving detection accuracy between 97.05% and 97.87%.
• Precision, recall, and F1-measure values were also notably high, ranging from 96.99% to 97.95%, 99.80% to 99.95%, and 98.45% to 98.87%, respectively.
• The refined feature set derived from ensemble selection significantly improved DL model performance in IoT attack detection.

Conclusions:

• The proposed ensemble feature selection combined with DL models offers a robust solution for detecting IoT network attacks.
• The findings highlight the effectiveness of this hybrid approach in improving cybersecurity for IoT environments.
• This research contributes to developing more secure and reliable IoT ecosystems through advanced threat detection.