Intelligent software defined networking: Long short term memory‐graded rated unit enabled block‐attack model to tackle distributed denial of service attacks

Abstract

AbstractSoftware defined networking (SDN) is the next‐generation network. SDN enhances the programming flexibility, speed and automation to improve the network's performance. In recent times SDN has played a vital role in networking technology. It communicates with underlying hardware infrastructure and directs traffic on a network. The most complicated issue in SDN is the control plane's single point of failure (SPF). The main reason for raising the SPF problem in SDN by distributed denial of service (DDoS) attacks. The network collapses during failures in SDN, and the control plane is considered a management controller. Therefore, a novel intrusion detection and prevention system (IDPS) is proposed in the proposed approach to address SDN's problems mentioned above. In the proposed approach, a long short term memory (LSTM) and graded rated unit (GRU) deep learning model is proposed as the “Block‐Attack” model. The main objective of using LSTM and GRU in the proposed approach is to enhance the rate of accuracy in detecting DDoS attacks in an SDN environment. The CICDDoS2019 dataset is used for experimental result analysis in the proposed approach. Initially, the dataset is fed into the preprocessing stage. Using the K‐medoid technique, raw datasets are preprocessed to reduce the model's sensitivity to low density. In the proposed approach, the support vector machine based machine learning (SVM‐ML) technique is utilized to prevent DDoS attacks in a Mininet‐based emulation. Then, LSTM and GRU deep learning (DL) techniques are used to define the block‐attack model to enhance the detection performance. The experimental results of the proposed approach “Block‐Attack” model attain 98.5% of accuracy to detect and prevent the DDoS attacks and 95.5% of accuracy for SVM based method.