A New Website Fingerprinting Method for Tor Hidden Service

Although anonymous communication systems protect user privacy, they also facilitate evasion of network censorship. Currently, evaders use anonymous hidden services to carry out various illegal activities, which pose a serious threat to network management. To address this problem, a new website fingerprinting method that combines the Bidirectional Encoder Representations from Transformers (BERT) model and a Long Short-Term Memory (LSTM) network was proposed to improve the accuracy of website fingerprinting for Tor hidden services. The proposed method uses the BERT model to extract the semantic features of webpage content, and the LSTM model is combined to capture the long-term dependencies to achieve efficient recognition of website fingerprints. This method not only deals with the homepage, but also considers the features of sub-pages in depth, which can effectively improve the performance of the model in closed-world and open-world scenarios through deep textual and time-series feature learning. The experimental results show that compared with existing state-of-the-art techniques, the proposed method exhibits better performance in terms of key performance metrics.