Hybrid Optimization Framework for Feature Selection in WSN Using Red Kite and Firefly Algorithms with Ensemble Learning for Intrusion Detection

Providing cybersecurity in Internet of Things (IoT) environments, especially within Wireless Sensor Networks (WSNs), is highly important for intrusion detection. A Hybrid Optimization Framework for Feature Selection, which brings together the Red Kite Optimization Algorithm (RKOA) and the Firefly Algorithm (FFLY), is proposed in this research to extract the most effective features with the least redundancy to detect the intrusion. For classification, an ensemble machine learning model involving Long Short-Term Memory (LSTM), Bidirectional LSTM (BiLSTM), and Bidirectional Gated Recurrent Unit (BiGRU) networks is employed to identify rich temporal patterns and boost detection capabilities. The advanced framework is benchmarked using the NSL-KDD dataset on various metrics that include accuracy, specificity, sensitivity, F1-score, and Area Under the Curve (AUC). Experimental findings show that the hybrid optimization method successfully identifies the optimal features, and the ensemble model successfully identifies different attacks. This solution not only provides improved detection accuracy but also guarantees scalability, providing a strong defense mechanism to protect IoT-based WSNs from changing cyber threats.