Automatic Smart Device Identification Based on Web Fingerprint and Neural Network

Cyberspace smart devices refer to the general term for devices that can connect to the Internet and have data processing functions. With the development of the Internet of Things technology, new smart devices such as webcams have brought convenience to people's lives, but also exposed many security vulnerabilities. Discovering smart devices in cyberspace is the prerequisite and basis for implementing cyberspace security management. However, the traditional pattern matching identification method requires manual extraction of device keywords, and keeping the keywords intact and updating the fingerprint database hinders accurate and large-scale device discovery. In this regard, this paper proposes a smart device identification method based on Web fingerprints and neural network. We use asynchronous stateless scanning and web crawlers to obtain the target's HTTP response data, extract the text in the response data based on natural language processing technology, and use neural networks to build a classification model. After processing, the response data of each IP is converted into concise text as a feature vector, and these texts are finally used to train the neural network model to realize the identification of smart devices. In order to prove the effectiveness of the algorithm, this paper uses Python to implement a prototype system and conduct experimental verification to evaluate its performance. Experiments show that among the four neural network models used, the RCNN model can converge in the shortest time and reach a training accuracy of 98.66%, and an accuracy of 90.59% on the test set. It is verified that the algorithm proposed in this paper is feasible and has a good ability to recognize smart devices.