An Automated Multi-Tab Website Fingerprinting Attack

Yin, Qilei; Liu, Zhuotao; Li, Qi; Wang, Tao; Wang, Qian; Shen, Chao; Xu, Yixiao

doi:10.1109/tdsc.2021.3104869

Cited by 21 publications

(3 citation statements)

References 22 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Xu et al [105] presented a new BalanceCascade-XGBoost scheme to identify the start point of the second page, they also developed a new classifier based on random forests which can accurately classify webpages given only the small chunk of packets between the start time of the two webpages. Yin et al [106] built a WF classifier based on XGBoost algorithm, replacing the original random forest classifier in [105]. Cui et al [107] proposed a splitting algorithm based on Hidden Markov Model to identify two continuous network traces and a sectioning algorithm to identify overlapping network traces, which divided the trace into multiple sections and performed website prediction on each section independently, based on the hypothesis that if two traces overlap, the beginning of the first trace and the end of the second trace would be unaffected.…”

Section: B Challenges and Solutionsmentioning

confidence: 99%

A Survey on Deep Learning for Website Fingerprinting Attacks and Defenses

Liu

2023

IEEE Access

View full text Add to dashboard Cite

The attacks and defenses on the information of which website pages are visited by users are important research subjects in the field of privacy enhancing technologies, they are termed as website fingerprinting (WF) attacks and defenses. Nowadays, deep learning is an important tool in many research areas, including WF attacks and defenses. In this paper, we offer a comprehensive survey on deep learning for WF attacks and defenses. After a brief introduction, we first summarize deep learning, WF attacks, and WF defenses. For deep learning, we review the common paradigms, architectures, and performance metrics. For WF attacks, we review the approaches, challenges and solutions. The approaches include deep learning, traditional machine learning, and other methods. Challenges and solutions cover multi-tab browsing, concept drift, and the base rate fallacy. For WF defenses, we review the strategies and approaches. Then, we survey deep learning for WF attacks, and deep learning for WF defenses. In deep learning for WF attacks, we survey in detail the deep learning paradigms, architectures of WF attack models, and the performance of several representative WF attack models, and look into the future. In deep learning for WF defenses, we survey the architecture, efficacy and overhead of deep learning models in WF defenses, and look into the future. In the end, we summarize this paper.

show abstract

Section: B Challenges and Solutionsmentioning

confidence: 99%

A Survey on Deep Learning for Website Fingerprinting Attacks and Defenses

Liu

2023

IEEE Access

View full text Add to dashboard Cite

show abstract

“…For instance, a user may browse two different websites simultaneously. So, the visiting of multiple websites at the same time inevitably brings in overlapping times [25,26]. Some previous works [25,26] have explored the issue of multiple websites browsing, and we will not discuss it further in this paper.…”

Section: Definitionsmentioning

confidence: 99%

A Practical Website Fingerprinting Attack via CNN-Based Transfer Learning

Pan,

Tang,

2023

Mathematics

View full text Add to dashboard Cite

Website fingerprinting attacks attempt to apply deep learning technology to identify websites corresponding to encrypted traffic data. Unfortunately, to the best of our knowledge, once the total number of encrypted traffic data becomes insufficient, the identification accuracy in most existing works will drop dramatically. This phenomenon grows worse because the statistical features of the encrypted traffic data are not always stable but irregularly varying in different time periods. Even a deep learning model requires good performance to capture the statistical features, its accuracy usually diminishes in a short period of time because the changes of the statistical features technically put the training and testing data into two non-identical distributions. In this paper, we first propose a convolutional neural network-based website fingerprinting attack (CWFA) scheme. This scheme integrates packet direction with the timing sequence from the encrypted traffic data to improve the accuracy of analysis as much as possible on few data samples. We then design a new fine-tuning mechanism for the CWFA (FM-CWFA) scheme based on transfer learning. This mechanism enables the proposed FM-CWFA scheme to support the changes in the statistical patterns. The experimental results in closed-world and open-world settings show that the effectiveness of the CWFA scheme is better than previous researches, with the slowest performance degradation when the number of data decreases, and the FM-CWFA scheme can remain effective when the statistical features change.

show abstract

“…In this case, the user's browsing objective would be the google.com image results web-page containing pictures of cats. Web browsing is a sequential task [21], [22]-even when a user desires to accomplish multiple tasks simultaneously, the user is limited by control over a single mouse cursor. Thus, an arbitrary web browsing objective can be broken down into a series of navigation steps, or interactions.…”

Section: Rlbrowse: Automated Web Browsingmentioning

confidence: 99%

RLBrowse: Generating Realistic Packet Traces with Reinforcement Learning

Griessel

Stephan

Mieth

et al. 2022

NOMS 2022-2022 IEEE/IFIP Network Operations and Management Symposium

View full text Add to dashboard Cite

Automated Web Browsing tools, such as Selenium and headless browsers, are used to collect traffic traces from networked applications, with which statistical models describing the traffic are obtained. However, we show that traces from Selenium and headless browsers have markedly different traffic characteristics than human generated traces, with potential impact on the quality of the obtained models. To overcome this limitation, we propose RLBROWSE, an automated web automation framework that imitates human browsing habits by separating web automation from the browser using reinforcement learning. By separating the browser and automation tool, RLBROWSE improves on 9 out of the 13 traffic trace features tested. The distribution of packet sizes in a trace improves the most, with a nearly 400 % improvement. We test RLBROWSE by collecting a corpus of network packet traces on a set of human-navigated website browsing sessions, and by RLBROWSE and Selenium.In the subsequent analysis, we identify key differences in the resulting packet traces.

show abstract

An Automated Multi-Tab Website Fingerprinting Attack

Cited by 21 publications

References 22 publications

A Survey on Deep Learning for Website Fingerprinting Attacks and Defenses

A Survey on Deep Learning for Website Fingerprinting Attacks and Defenses

A Practical Website Fingerprinting Attack via CNN-Based Transfer Learning

RLBrowse: Generating Realistic Packet Traces with Reinforcement Learning

Contact Info

Product

Resources

About