BA-BNN: Detect LDoS Attacks in SDN Based on Bat Algorithm and BP Neural Network
Xinmeng Li, Nengguang Luo, Dan Tang, Zhiqing Zheng, Zheng Kun Qin, Xinxiang Gao · 2021
Low-rate Denial of Service (LDoS) attack is a novel form of the denial-of-service attack. Consequently, there is something difficult to detect it by conventional attack detection techniques. BP neural network detects LDoS attacks is feasible because of its features such as high non-linear. However, it also has the defects such as poor global search ability. Bat algorithm that has the strong global search ability can make up those defects. We proposed a LDoS attack detection method that is on account of Bat Algorithm and BP Neural Network (BA-BNN). For the sake of making a verification on how valid our method is, a series of LDoS attack experiments were carried out in a virtual Software Defined Network (SDN) composed of Mininet emulator and Ryu controller, obtaining the TCP and UDP traffic data. We use the algorithm called random forest to select characteristics of training data for BP neural network ameliorated by an intelligent algorithm, Bat algorithm, and construct the LDoS attack detection model. We compare our method with some other methods. The results show that the method we adopt is a highly accurate LDoS attack detection method. It has higher correct rate and detection rate, and lower false negative rate and false positive rate.