A Novel Framework for DDoS Attacks Detection Using Hybrid LSTM Techniques

The recent development of cloud computing offers various services on demand for organization and individual users, such as storage, shared space, networking, etc. Although Cloud Computing provides advantages it remains vulnerable to many types attacks that attract cyber criminals. Distributed Denial Service (DDoS) is the most common type attack computing. Consequently, professionals security experts have focused growth preventive processes towards DDoS attacks. Since become increasingly widespread, becomes difficult some methods based network flow features distinguish Further, monitoring pattern traffic changes accurate detection are important urgent. In this research work, deep belief feature extraction Hybrid Long Short-Term Memory (LSTM) model been proposed with NSL-KDD dataset. LSTM method, Particle Swarm Optimization (PSO) technique, which combined optimize weights neural network, reduces prediction error. This method used extract IP packets, identifies PSO-LSTM model. Moreover, accurately predicts normal detects anomalies resulting from architecture outperforms classification techniques including standard Support Vector Machine (SVM) in terms performance along results measurement accuracy, recall, f-measure, precision.