Barrier Number Estimation with Machine Learning for Intrusion Detection in Wireless Sensor Networks

Abstract

Intrusion detection in wireless sensor networks is crucial for ensuring network security. This study focuses on the problem of estimating the number of barriers necessary for effective intrusion detection in WSNs. The aim is to make accurate predictions to improve security optimization in WSNs. To this end, various regression models (Linear Regression, Ridge and Lasso Regression, Random Forest, Support Vector and Gradient Boosting) were applied on a dataset including parameters such as field size, sensing range, transmission range, and the number of sensor nodes. The performance of the models was evaluated with metrics such as R2, RMSE, MAE, and MSE, and validated with 5-fold cross-validation. The results show that the Linear Regression model achieved the best performance with the lowest error values (RMSE 0.0181, MAE 0.0136, and MSE 0.0003), followed closely by Ridge Regression. These findings highlight the effectiveness of simple linear models in accurately predicting barrier requirements, supporting the optimization of WSN security systems