•  
  •  
 

Abstract

Intrusion Detection Systems (IDS) are crucial for safeguarding Internet of Things (IoT) networks against security threats. The integration of Support Vector Machine (SVM) with smart IDs has been a significant advancement in detecting anomalous activities. This research contributes to this field by implementing the Gaussian Pyramid (GP) algorithm, significantly reducing the processing amount and storage space required for large IoT network traffic datasets. This approach enables the GP model to classify thousands of data points in large-scale problems with high-dimensional input space. Notably, the GP model, with various kernel sizes, outperforms traditional nonlinear SVM and Artificial Neural Networks (ANN) in terms of efficiency and accuracy. For instance, with kernel sizes of 5, 7, and 9, the GP model demonstrated superior performance on the NSL-KDD dataset, achieving accuracy and AUC (Area Under the Curve) values higher than both nonlinear SVM and ANN. In kernel size 9, the GP model achieved the highest overall accuracy of 0.96% on the CIC-DDoS2019 dataset. The experimental results confirm that applying the GP model to IoT data traffic significantly reduces time complexity and enhances the performance of binary and multi class SVM, marking a substantial advancement in IoT intrusion detection.

Keywords

Gaussian Pyramid, GP Model, IDS, IoT, SVM

Subject Area

Computer Science

Article Type

Article

First Page

1714

Last Page

1724

Creative Commons License

Creative Commons Attribution 4.0 International License
This work is licensed under a Creative Commons Attribution 4.0 International License.

Share

COinS