Effective Denial‐of‐Service Attack (DoS) Detection Using Progressive Cyclical Convolutional Neural Network (CNN) in Wireless Sensor Networks

Wireless sensor networks (WSNs) are vital for modern applications but remain highly vulnerable to denial‐of‐service (DoS) attacks because of their limited resources. To address this, this paper proposes a DOSAD‐WSN‐PCCNN, an effective and lightweight DoS detection framework based on progressive cyclical convolutional neural network (PCCNN). This approach integrates advanced preprocessing using variational Bayesian–founded extreme correntropy cubature Kalman filter (VBECKF), feature selection using variable velocity strategy particle swarm optimization algorithm (VSPSOA), and parameter optimization using Tyrannosaurus optimization algorithm (TOA). This hybrid strategy significantly enhances detection accuracy, efficiency, and robustness. Experimental results on the WSN‐DS dataset demonstrate up to 99.5% accuracy and outperform baseline methods by over 25% in key metrics. The proposed system shows strong potential for real‐time deployment in resource‐constrained WSN environments, significantly improving network resilience against evolving DoS threats.