A State-of-the-Art Intrusion Detection Framework for Wireless Sensor Networks Integrating Cross-Stage Partial Recurrent Networks

Wireless Sensor Networks (WSNs) are highly vulnerable to security threats due to their distributed and resource-constrained nature, making effective intrusion detection essential. This study proposes a robust intrusion detection system designed to achieve high accuracy with low computational complexity. The performance of the proposed model is tested on benchmarked datasets such as WSN-DS, NSL-KDD, CICIDS-2017, and KDDCup99. First, Min–Max normalization is used to scale the input variables, and the Synthetic Minority Over-sampling Technique (SMOTE) is then employed to solve class imbalance problems. For feature selection, the Enhanced Parrot Optimization (EnPaO) method is used along with chaotic Chebyshev mapping to enhance convergence and improve the quality of selection. After optimization, the selected features are passed through the suggested cross-stage partial recurrent network (CrPR_Net). This model is capable of capturing spatial and temporal patterns from network traffic. Also, the Addax Optimization (AdO) algorithm is employed to optimize the model’s hyperparameters, improving its performance. The performance of the proposed EnPaO + CrPR_Net architecture is tested in Python by using different performance metrics such as accuracy, precision, recall, specificity, F1-score, and mean square error. The model provides excellent results with an accuracy of 99.91% and a negligible error rate. Experimental results demonstrate that the proposed model is effective and reliable for detecting a wide range of intrusion types across the evaluated benchmark WSN datasets.