Intrusion Detection System in Wireless Sensor Networks using Modified Recurrent Neural Network with Long Short-Term Memory

Leveraging deep learning, Intrusion Detection Systems (IDS) for Wireless Sensor Networks (WSN) enhance security through advanced analytics, detecting and mitigating network intrusions in real-time for robust protection. Evaluations and scalability considerations are critical for improving the applicability and robustness of intrusion detection systems based on deep learning techniques. This paper introduces a novel approach: a modified Recurrent Neural Network with Long Short-Term Memory (RNN-LSTM) for IDS in WSN classification, utilizing the NSL-KDD dataset. Rigorous performance validation demonstrates outstanding metrics, including an accuracy of 99.95%, precision at 99.93%, and a robust recall of 95.69%, with the F1-score reaching 99.80%, surpassing benchmarks set by existing methods. Notably, current approaches like Deep Neural Network (DNN), Convolutional Neural Network-LSTM (CNN-LSTM), Conditional Generative Adversarial Network (CGAN), and Stacked Autoencoders-DNN (SAE-DNN) face limitations in achieving optimal accuracy and robustness. The proposed RNN-LSTM method exhibits significant advancements in addressing these challenges for effective intrusion detection in WSN