Low-rate Denial of Service (LDoS) attack exploits security vulnerabilities of network protocols adaptive mecha- nisms to launch periodic bursts. It poses severe destructiveness to the quality of service of TCP applications. Therefore, the im- plementation of the detection of LDoS attacks is a concern among scientific communities. But the existing coarse-grained detection methods have poor detection performance and adaptability. To achieve accurate detection of LDoS attacks, the adaptive Kohonen network based fine-grained detection (AKN-FGD) of LDoS attacks is proposed. Based on the burstiness and periodicity characters of attack traffic, the Smith-Waterman (SW) algorithm is used to estimate the pulse period that is the length of the detection unit. Subsequently, cluster analysis is performed for each detection unit using the adaptive Kohonen network (AKN) algorithm according to the discreteness of traffic suffering from LDoS attacks is more pronounced than that of legitimate traffic. Finally, the existence of LDoS attacks can be verified in view of a novel decision metric, denoted as the anomaly degree based on the clustering results.
Low-rate denial of service (LDoS) attacks reduce the quality of network service by sending periodical packet bursts to the bottleneck routers. It is difficult to detect by counter-DoS mechanisms due to its stealthy and low average attack traffic behavior. In this paper, we propose an anomaly detection method based on adaptive fusion of multiple features (MAF-ADM) for LDoS attacks. This study is based on the fact that the time-frequency joint distribution of the legitimate transmission control protocol (TCP) traffic would be changed under LDoS attacks. Several statistical metrics of the time-frequency joint distribution are chosen to generate isolation trees, which can simultaneously reflect the anomalies in time domain and frequency domain. Then we calculate anomaly score by fusing the results of all isolation trees according to their ability to isolate samples containing LDoS attacks. Finally, the anomaly score is smoothed by weighted moving average algorithm to avoid errors caused by noise in the network. Experimental results of Network Simulator 2 (NS2), testbed, and public datasets (WIDE2018 and LBNL) demonstrate that this method does detect LDoS attacks effectively with lower false negative rate.