Detection of IoT Botnet Using Recurrent Neural Network

Abstract

AbstractThe Internet of Things (IoT) is one of the most used technologies nowadays. Hence, the number of DDoS attacks generated using IoT devices has raised. Normal anomaly detection methods, like signature-based and flow-based methods, cannot be used for detecting IOT anomalies as the user interface in the IOT is incorrect or helpless. This paper proposes a solution for detecting the botnet activity within IoT devices and networks. Deep learning is currently a prominent technique used to detect attacks on the Internet. Hence, we developed a botnet detection model based on a bidirectional gated recurrent unit (BGRU).The developed BGRU detection model is compared with gated recurrent unit (GRU) for detecting four attack vectors Mirai, UDP, ACK, and DNS generated by the Mirai malware botnet, and evaluated for loss and accuracy. The dataset used for the evaluation is the traffic data created using the Mirai malware attack performed on a target server using C&C and scan server.KeywordsInternet of Things (IoT)BotnetGated recurrent unit (GRU)Bidirectional gated recurrent unit (BGRU)Deep learning