Detection and Prevention of Distributed Denial of Service Attack to IoT servers using Recurrent Neural Network with Long Short Term Memory

The Internet of Things' (IoT) explosive growth has created serious security issues, with Distributed Denial of Service (DDoS) assaults becoming a serious danger to the dependability and availability of IoT servers. Using a Recurrent Neural Network (RNN) with Long Short-Term Memory (LSTM) units, this research suggests an intelligent DDoS detection and prevention framework that is especially made to capture the temporal patterns of Internet of Things traffic. The model uses a Softmax-based classifier and a many-to-one LSTM architecture to accurately identify changing attack behaviors. Robust data handling is ensured by preprocessing methods such as Min-Max normalization and K-Nearest Neighbors (KNN) imputation. By examining traffic characteristics like packet rates and IP address mapping, the architecture incorporates attack prevention measures. The model's efficacy is demonstrated experimentally using the benchmark BoT-IoT dataset, which yields better detection performance and real-time responsiveness than traditional CNN and ensemble-based methods. The outcomes validate the model's capacity for proactive and flexible DDoS protection in Internet of Things settings. The Internet of Things' (IoT) explosive growth has created serious security issues, with Distributed Denial of Service (DDoS) assaults becoming a serious danger to the dependability and availability of IoT servers. Performance analysis of the proposed model represents robustness and accuracy of 98.7% to Sybil attack, 98.6% to sink hole attack, 98.1 to Syn flood attack and 99.1 to UDP flood attack against state of art approaches.