| With the rapid development of IoT-related technologies in recent years,it has facilitated people’s daily work and life.However,due to the lack of computing resources,insufficient storage resources,shortage of manual supervision of IoT devices,the security technology applied to the device is not well adapted to the speed of its development and some other reasons,there are a large number of vulnerable IoT devices,which increased the security threat of Botnet attacks based on the IoT with the explosive growth of IoT devices.Therefore,it is important for research on IoT Botnet detection.This paper summarizes the characteristics of IoT Botnets and related detection methods,and conducts in-depth research on the basis of IoT Botnet traffic analysis and machine learning related technologies.The main innovations of this paper are summarized as follow:(1)The Mirai,a typical representative of the IoT Botnet,is used for a example.Analyzing its main components and building process from the source code level,and deeply understand and analyze the characteristics of the IoT Botnet through the establishment of the Mirai Botnet,and compare IoT Botnet traffic with normal traffic for the foundation of subsequent detection.(2)An IoT Botnet detection model based on RF-RFECV and Light GBM is proposed.Firstly,the RF-RFECV feature selection algorithm and the theoretical basis of Light GBM are introduced,and then the flow feature extraction,data preprocessing,feature extraction and classification detection processes of the model are introduced.Finally,experiments show that the model is effective for IoT Botnets Detection.(3)An IoT Botnet detection model based on unsupervised ensemble learning is proposed.Firstly,the JL projection and the basic theory of outliers based on ensemble learning are introduced,and then the data preprocessing,base classification model selection and outlier score combination function method are introduced in detail according to the specific architecture of the model.Finally,the experimental results show that the model has applicability on different devices,and all have high detection rate and low false alarm rate.Facing the current growing Internet of Things Botnet,the detection model proposed in this paper can effectively detect IoT Botnet.This has great practical significance for the security protection of IoT devices and the protection of the network security of IoT. |
