Security Analysis Of IoT Pairing Process And Traffic Identification

With the smart home devices entering into people’s daily life gradually,the security of Io T devices has received extensive attention from researchers.The process of device pairing is very important for the secure access and authorization of smart home devices.Currently,there is a lack of relatively complete analysis of the communication security during the pairing process of Io T devices,and there is a lack of effective monitoring for traffic identification during the pairing process.This paper conducts research on the above issues.Specific research work and contributions including:(1)A simulated smart home platform is designed and built,which collects the pairing process traffics of a total of 15 Io T devices connected by Wi-Fi from multiple smart home manufacturers,and constructs a traffic dataset of device pairing process with domestic mainstream smart home devices.(2)A state transition model is used to describe the pairing process of smart home devices,and a security analysis of the communication process between devices,mobile phone Applications and cloud servers in the smart home platform is carried out.The process of device pairing is divided into four stages: device discovery,device networking,remote authentication and remote binding,and then the defects in different stages and the security problems caused by the defects are analyzed.The defects in different stages and the security problems caused by the defects are analyzed,and experimental analysis is carried out on multiple real devices.(3)A method is proposed to identify the pairing process in the traffic and identify the paired devices according to the statistical characteristics of the traffic.The method utilizes the statistical characteristics of the network traffic,and then uses a variety of machine learning classification models to classify the device communication traffic in the pairing process.The test results show that the method can effectively identify the device type,and the random forest algorithm can achieve an accuracy of 89.48% on the public dataset.