Internet Of Things Device Identification Technology Based On Message Semantic Information Extraction

With the wide application of Internet of things devices in production and life,more and more Internet of things devices are connected to the cyberspace.However,due to its own reasons and manufacturers'reasons,there are a lot of loopholes in IoT devices.The major vulnerability platforms around the world will publish the vulnerability information and the affected IoT equipment information from time to time.Effective identification of Internet of things devices in cyberspace can make use of these public vulnerability information to patch Internet of things devices,thereby improving the security level of Internet of things devices and safeguarding cyberspace security.However,there are two major problems in the identification of IoT devices:utilizing protocol packets containing semantic information of IoT devices to identify IoT devices;identifying Internet of things devices with protocol messages that lacking semantic information or using irregular format.Regarding the problem of utilizing protocol messages containing semantic information of IoT devices to identify IoT devices,this paper proposes a method to extract semantic information of IoT devices to solve this problem.The semantic information extraction method of IoT devices is to identify IoT devices by using named entity recognition.It has a finer granularity and higher accuracy compared with the traditional identification methods.The experimental results show that the precision rate of this method is 95.49%,the recall rate is 94.55%,and the F₁ score is 95.02%.Regarding the problem of identifying Internet of things devices with protocol messages that lacking semantic information or using irregular format,this paper proposes an unstructured information extraction method of IoT devices to solve this problem.The unstructured information extraction method of Internet of things equipment is to identify these protocol messages by combining manual sampling verification and automatic identification.Firstly,the feature of unstructured information in protocol message is extracted.Next,the feature of unstructured information in protocol message is clustered by the method of agglomerative clustering.After that,the random sampling method is used to sample from the cluster.Finally,the classification model is trained by using the characteristics of the unstructured information in the marked protocol message to realize the automatic identification of IoT devices.The experimental results show that the precision rate of this method is 93.83%,the recall rate is 93.50%,and the F₁ score is 93.53%for those IoT devices which lack semantic information or whose format is not standardized.This paper utilized the above methods to identify IoT devices,and establishes a fingerprint database of 1228 kinds of different IoT devices,including 188 manufacturers and 17 types of devices.