Generating Similar Traffic With Rendering Equivalent Objects Against Website Fingerprinting Attacks

When visiting a web page with a browser,a web trace is generated,and a web trace records a series of requests and responses when visiting a web page.Including visiting some special pages,such as hospitals,schools,banks,etc.,it is also because of this that there is a risk that web traces will record and leak personal privacy.To guard against the risk of privacy leaks,technicians began encrypting content transmitted across the network.For example,more and more websites have begun to use the HTTPS protocol.However,under the existing technical means,even if the transmitted traffic is encrypted,the traffic analysis attack can still easily identify the page that the user is visiting after analyzing over to the classifier,which provides convenience for some network attackers to obtain user privacy information.This paper proposes a new way to defend against traffic analysis attacks,SOST,which fills the size of web objects on the server side without changing the rendering effect of the browser.At the same time,we propose the concept of rendering equivalent,we modify the object properties in the original page,so that the rendering results of the modified page and the original page are the same,without affecting the availability of other properties of the page.Rendering equivalent is achieved by adding methods that make the object sequence of two pages the same size and order,empty objects,and filled objects.Inspired by the idea of kanonymity algorithm,in a set of private web pages,using similar website fingerprinting strategies to confuse the analysis and identification of classifier web pages can achieve good results.More importantly,after applying this strategy,the availability of relevant statistics still supports statistical analysis of data.Finally,relevant experiments are also done to confirm the feasibility of the ideas in this paper.