Optimized Hybrid Deep Learning Approach for Detecting DoS Attacks Using Feature Selection in Wireless Sensor Networks

The rapid growth of smart cities, healthcare monitoring, and environmental sensing relies heavily on the real-time data processing capabilities of Wireless Sensor Networks (WSNs). However, these networks face significant security challenges, particularly from malicious activities like Denial of Service(DoS) attacks, which can disrupt critical operations. A hybrid intrusion detection model that combines Convolutional Neural Networks (CNN) and Deep Neural Networks (DNN) to improve the detection of such threats has been proposed. The preprocessing phase includes the Synthetic Minority Over-sampling Technique (SMOTE) which mitigates class imbalances ensuring robust performance across diverse attack scenarios. The model incorporates advanced feature selection techniques including SelectKBest and Grey Wolf Optimization (GWO) to identify the most critical features for enhanced detection accuracy. The CNN component effectively captures local traffic patterns, which are then refined by the DNN for accurate classification. This hybrid model's effectiveness in detecting diverse DoS attack scenarios is evaluated using WSN-DS dataset, ensuring robust security for WSNs. By achieving an impressive 98% detection accuracy, this hybrid model proves to be effective in intrusion detection. Its applications can significantly strengthen the security and reliability of WSNs, ensuring uninterrupted operations in key sectors such as smart cities, healthcare, and environmental monitoring.