Detecting Distributed Denial of Service Attacks in Software-Defined Networks with a Deep Learning Approach
Subject Areas : AI and Roboticsیونس مهدی زاده 1 * , مهدی صادق زاده 2
1 -
2 -
Keywords:
Abstract :
The growth of cloud computing has led to the development of software-defined networks. These networks enable dynamic management and performance improvement. Security threats in this type of network are a growing concern. Especially, the controller of these networks is an attractive target for hackers and distributed denial of service attacks. Many researchers have proposed different methods to detect these attacks, whose false detection rate is very high and has led to a decrease in detection accuracy. For this purpose, in this research, the focus is on detecting distributed denial of service attacks through deep learning using prominent features of packets. After pre-processing and preparing the data, the proposed method separates the salient and important features of the packages through the support vector machine method and finally by using an innovative hybrid neural network consisting of convolutional neural network, sample recurrent neural network and Long-term short-term memory neural network separates attack packets from normal packets. A standard data set has been used to evaluate the proposed method through standard evaluation criteria such as detection accuracy, precision, false detection rate and harmonic mean accuracy. The findings show that the proposed method detects distributed denial of service attacks with 95.2% detection accuracy, 92.09% precision, 5.1% false alarm rate, and 93.87% F1_measure.