DDoS Attack Detection Algorithm Based on Hybrid Traffic Prediction Model

Abstract

In order to detect Distributed Denial of Service (DDoS) attack accurately and rapidly, a DDoS attack detection algorithm based on hybrid traffic prediction model (DADA- HTPM) is proposed. Firstly, the hybrid traffic prediction model (HTPM) is proposed, in which the local projection, phase space reconstruction and RBF neural network technology are all used to restore the chaos of network traffic and train the network traffic samples to predict the future network traffic accurately. Then, the DDoS attack detection algorithm is developed based on two thresholds to reduce the false alerts caused by environment noise. Simulation results show that the DDoS attack detection algorithm based on hybrid traffic prediction model presented in this paper not only has higher traffic prediction accuracy and lower complexity, but also can detect DDoS attack quickly and accurately.