RETRACTED ARTICLE: Detection of distributed denial of service using deep learning neural network

Abstract

The need for developing a neural network classifier in an intrusion detection system for network security purpose is a necessary. Today, worldwide various types of sophisticated attacks damage the network in both wire and wireless. The medium of wireless network is air which is used to transmit the data where several categories of attacks damage the network system. Among the attacks, denial-of-service (DoS) attack easily access the network and also very difficult to prevent. Therefore, the protection of various resources in the network is a challenging one and the detection of DoS attack process is an important issue in the network. For this purpose, it requires high-performed machine learning classifier with less computational time, minimum false positive and high detection accuracy. This paper evaluates the network performance using deep learning neural network classifier with cost minimization strategy for a publicly available dataset. The proposed approach utilizes the KDD Cup, DARPA 1999, DARPA 2000, and CONFICKER datasets. The performance metrics such as detection accuracy, cost per sample, average delay, packet loss, overhead, packet delivery ratio and throughput are used for the performance analysis. From the simulation result observed that DNN Cost minimization algorithm provides better result in terms of high detection accuracy 99% with less false reduction, high average delay, less packet loss, less overhead, high in packet delivery ratio and throughput is high compared to existing algorithm.