Automatic Discovery And Extraction Of IoT Device Organization Information At Large-scale

With the development of Internet technology,the Internet of Things(IoT)technology is also developing at a high speed and high quality.On one hand,the Internet of Things has increased the connection between things and things,things and people,strengthening the relationship between people,and further realized the communication between all things.It accelerates the advancement of technology.On the other hand,the Internet of Things is simple,convenient and energy-saving.It is an important technology for achieving high efficiency and high quality in the new century.However,security incidents such as IoT attacks in recent years have affected the development of the network and the security of the country.IoT technology has gradually attracted the attention of all walks of life,and it has become a research hotspot in academic area.However,although many researchers have studied various aspects of the IoT,in the research literature on security and management,it is still found that a large number of IoT devices can be accessed and visible through IP addresses,which is extremely likely to cause critical facilities and information is leaked and misused,causing cybersecurity incidents.Therefore,we conduct a more in-depth study of the attribute information of IoT devices from a security perspective.We exploit the organization information extraction technology of large-scale IoT devices,including IoT device identification and IoT device organization information extraction.(1)In the IoT device identification,we use the difference of application layer services to extract the content of the application layer response data as the characteristics of IoT devices,and combines the four conventional machine learning classification algorithms to improve the precision of IoT device recognition.(2)In the IoT device organization information extraction,through practice observation,we find that many device organizers will write relevant details to the application layer services on the IoT device,such as the company or user name,etc.The hard-coded information can be used as the basis for device organization information extraction.Therefore,we propose two information extraction technology based on rule and named entity recognition to obtain device organization information in application layer response packets.In order to verify the feasibility of device identification and organization information extraction,thesis builds a prototype system and carries out experimental verification.The experimental results show that the machine learning classification algorithm has good recognition performance for IoT devices,the precision is 97%and coverage is 97%.Thesis collected 93 million HTTP response data packets and found nearly 4.3 million IoT devices exposed on the Internet through device identification technology.In addition,thesis has extracted more than 600,000 organization names of IoT devices.The results show that the IoT device identification and organization information extraction technology has essential value for the security and defense of cyberspace.