Research On IoT Device Identification Technology Based On Active Detection

The increasing popularity of IoT devices has brought about significant network security issues,as the disparity between the speed of device proliferation and security measures has made devices vulnerable to attacks and privacy breaches.Through the application of host detection and identification technology,the equipment can be effectively identified and safely managed,but current methods rely on manual labeling and are inefficient,making them unsuitable for the diverse and heterogeneous device control requirements of the IoT.In this paper,in order to efficiently and rapidly identify IoT devices in cyberspace and to carry out effective network security management,IoT device identification techniques are investigated.Currently,there are challenges such as the multitude of response data formats.The active detection and identification method of IoT devices is used to address these challenges in this paper.This method comprehensively uses lightweight machine learning models to mine the structured and unstructured protocol features of IoT devices,and constructs a feature library covering mainstream devices.Combined with the Django framework,a prototype system for IoT device identification is designed and implemented,and it is verified in a real network environment.The results show that the method proposed in this paper is actually effective.The main contributions of this paper are:(1)Proposing a clustering algorithm based on structured information features of IoT devices.By extracting multi-dimensional features of devices based on response message header fields,using principal component analysis for feature dimensionality reduction,and analyzing data density distribution using k-dist graph,the paper achieves high-quality clustering results using DBSCAN and establishes a correspondence between 24,195 features and device types and brand manufacturers,including five types of IoT devices such as webcam,router,printer,NVR,and DVR,and 14 mainstream brand manufacturers.(2)Proposing a clustering algorithm based on unstructured information features of IoT devices.By extracting text features from response data using N(3)-gram word segmentation combined with hash algorithms,and using DBSCAN for clustering,the paper achieves highquality clustering results and establishes a correspondence between 2,786 features and device information,including three types of IoT devices such as webcam,router,gateway,and 13 brand manufacturers.(3)Design and implement a prototype system for IoT device identification.The Django framework is used to design and realize the prototype system of IoT device identification,and the feature library of connected IoT devices is constructed based on feature mining and clustering.The system uses the cosine similarity calculation method to match and identify the devices under test.The prototype system was tested and verified in the IoT simulation laboratory.It took 10.95 s to identify 12 IoT devices,and the type identification accuracy rate was 100%.This paper has made a certain contribution to the active detection and identification of IoT devices,providing a reference for efficient and professional identification of IoT devices in real business scenarios.