DDoS attack detection in Internet of Things using recurrent neural network

Abstract

The Internet of Things (IoT) is an emerging communication paradigm due to its wide range of applications. For IoT, distributed denial of service (DDoS) attacks are becoming increasingly widespread, and the solutions to combat these attacks are in increased demand. The significant contributions of this paper include offering a novel algorithm named DALCNN (Detecting Attack using Live Capture Neural Network) for detecting DDoS attacks in IoT using the concept of recurrent neural network and implementation of a Software-defined-Network (SDN) using OpenDayLight platform. Furthermore, a three-tier architecture is proposed to classify and detect DDoS attacks. The algorithm classifies the kind of attack using a novel activation function and the machine/deep learning concepts. The proposed classifier is tested on 177 instances. The Simulation was carried out using the tools such as Mininet, Wireshark to generate the DDoS attack and accurately detect the different types of DDoS attacks in the network. The simulation results reveal that the values of benchmark parameters – accuracy, true positive rate, false-positive rate, precision, recall, F-measure, and ROC-area are: 99.98%, 0.999, 0.01, 0.999, 0.999, 0.999, and 0.999, respectively. Apart from it, the network performance evaluation of the various open-source controllers’ including Floodlight, Ryu, ONOS, and the OpenDayLight based on the performance metrics – Throughput, Latency, and Aggregate Controller performance as well was carried out. The Simulation results reveal that the proposed algorithm performs efficiently compared to the other existing algorithms. The proposed OpenDayLight controller performs way better than the other open-source controllers.