Detection of reduction-of-quality DDoS attacks using Fuzzy Logic and machine learning algorithms

Distributed Denial of Service (DDoS) attacks are still among the most dangerous on Internet. With advance methods for detecting and mitigating these attacks, crackers have improved their skills in creating new DDoS attack types with aim mimicking normal traffic behavior therefore becoming silently powerful. Among advanced types, so-called low-rate DoS at keeping a low level network traffic. In this paper, we study one techniques, called Reduction Quality (RoQ) attack. To investigate detection type attack, evaluate compare use four machine learning algorithms: Multi-Layer Perceptron (MLP) neural backpropagation, K-Nearest Neighbors (K-NN), Support Vector Machine (SVM) Multinomial Naive Bayes (MNB). We also propose an approach kind based three methods: Fuzzy Logic (FL), MLP Euclidean Distance (ED). FL, ED to above algorithms using both emulated real traces. show that Learning algorithms, best classification results obtained MLP, which, traffic, leads F1-score 98.04% 99.30% legitimate while, it 99.87% 99.95% Regarding EC, 98.80% 99.60% 100% However, better performance is cost larger execution time, since required 0.74 ms 0.87 datasets, respectively, where as 11’46” 46’48” classify respectively.