Session-based IoT Device Classification

With the rapid increase of Internet of things devices,it is necessary to understand which devices are connected in the Internet of things and whether they are running normally.The classification of Internet of things equipment can make operators plan and distinguish the Internet of things equipment,and can effectively monitor whether the Internet of things equipment is running normally,so as to timely deal with and replace the fault equipment,which greatly saves time and labor costs,and improves work efficiency.The classification of IOT devices is the premise of IOT intrusion detection and anomaly analysis.Only the accurate classification of IOT devices can effectively protect a certain type of devices.Based on the above requirements,this paper studies how to accurately classify IOT devices,and designs a session based IOT device classification architecture(1)A session based device classification model architecture for Internet of things is designed.The whole model architecture is divided into IOT device classification module and customer use module.The structure and function of each part of the module are introduced in detail.The IOT device classification module is composed of three parts:data collection,traffic imaging algorithm and model training/updating.In the data collection part,the collected data sets include the open source data sets of IOT devices shared by the network and the cooperation data sets The company collected the Internet of things device traffic,built convolution neural network model,and trained and tested it.The time stamp scanning algorithm and result display part are designed in the customer module to meet the actual needs of customers.(2)In this paper,a traffic mapping algorithm is proposed to deal with the traffic of Internet of things devices.Due to the irrationality and complexity of the features of IOT devices used in the current research,this paper proposes to split the IOT device traffic into sessions for processing,and then transform the session into gray image after cutting/filling and dimension transformation to process the IOT device data.This method includes four parts:traffic segmentation,traffic processing,image generation and IDX conversion.Thus,all kinds of Internet of things device traffic can be generated into a group of gray images.In these generated gray images of Internet of things devices,different gray images of Internet of things devices are divided into different categories and stored in the gray image library.These gray images can be used as the portraits of Internet of things devices by the traffic of Internet of things devices,which can be analyzed and identified by experts and displayed to users.In addition,in this study,the proposed traffic mapping algorithm is applied to other mainstream machine learning algorithms and compared.The data of the traffic mapping algorithm designed in this paper and the data without traffic mapping algorithm are input into the mainstream machine learning algorithm model for comparison,and the accuracy rate,recall rate and F1 value are used for comparative analysis.It can be seen that the proposed traffic mapping algorithm can greatly improve the accuracy of other algorithms in the scene of Internet of things device classification,and can greatly reduce the cost The time complexity of the algorithm simplifies the process of feature extraction.(3)The data set is divided into training set and test set.After training and testing,the classification accuracy of IOT devices in the data set based on the session classification model is more than 97%,which can accurately distinguish a variety of IOT devices,such as cameras,motion sensors,power switches,charging piles and medical devices.The proposed traffic mapping algorithm and session based IOT device classification model architecture are used in Beijing Yuanan IOT Technology Co.,Ltd.It effectively solves the problems existing in the Internet of things,such as the equipment data is huge and the types are messy,which can not be unified management and so on,and lays the foundation for the security protection of the Internet of things.