Abstract
To address the limitations of traditional machine learning method in the detection of minortiy class samples and the overall detection accuracy in Wireless Sensor Networks (WSN) intrusion detection, this paper proposes an intrusion detection method for WSN based ST-IAOA-XGBoost. A synthetic minority oversampling technique, Tomek Link (SMOTE-Tomek), is used to reduce the impact of data imbalance on detection performance. In the proposed IAOA, LTOBL is incorporated to generate an efficient and even distribution initial population during the optimization process. Nonlinear inertia weights and Gaussian distribution are introduced to enhance the global and local search capabilities. Lévy flight is incorporated to randomly perturb the optimal individuals, enhancing population diversity and preventing the search from falling into the local optimal solution. To verify the effectiveness of the proposed method, we use 11 benchmark test functions and make simulation experiments through the WSN-DS dataset and KDDCUP 99 dataset. Simulation results show that, compared with reference algorithms, the proposed IAOA performs well in the benchmark function test, the ST-IAOA-XGBoost achieved significant improvements in terms of Accuracy, Precision, Recall, and F1 score on the WSN-DS dataset, with increases of up to 17%, 12%, 17%, and 27%, respectively. Similarly, on the KDDCUP 99 dataset, the enhancements were up to 22%, 23%, 22%, and 23% for Accuracy, Precision, Recall, and F1 score, respectively.
Funding Information
  • Fundamental Research Funds for the Central University (3122018C022)
  • Key Program of National Natural Science Foundation of China (U2033210)

This publication has 39 references indexed in Scilit: