| In recent years,with the development of mobile communication technology,smart mobile devices have gradually become the principal method for users to access the Internet.More and more companies are becoming aware of the importance and urgency of mobile marketing,which promotes the rapid development of mobile advertising business.For the better accuracy of advertising and user recommendation,browser fingerprinting has gradually become the favor of mobile advertisers and the focus of the academia and industry.However,some advertisers “cooperate” with each other to implement cross-domain user linking,and then analyze users' behavior and preferences,which violates users' privacy to a great extent.Therefore,detection and defense of browser fingerprinting is an urgent problem to be solved.The existing detection of browser fingerprinting just uses simple rule matching for specific objects' or attributes' acquisition methods to identify browser fingerprinting,which is lack of universality.In addition,most of the current research work on the defense of browser fingerprinting directly prohibit device identification and are incompatible with the ecosystem of mobile Internet.In view of the above problems,this thesis studies the automatic detection and defense of mobile browser fingerprinting based user tracking,of which the main work includes the following three aspects.Firstly,this thesis studies the automatic detection of browser fingerprinting.Proceeding from the two aspects,i.e.,tracking method and tracking effect,automatic identification based on sensitive script call detection and page content association analysis is proposed.The former monitors the call of sensitive JavaScript APIs and uses machine learning to detect browser fingerprinting.The latter identifies web tracking by analyzing the relevance of page content and user operations.Furthermore,the validity of the method is verified with the tracking websites that have been published and the normal Alexa ranking websites.Secondly,this thesis studies the defense of browser fingerprinting.For the browser fingerprint features with high rate of identification and usage,different noise is added to the corresponding return value of the function based on the different properties' acquisition methods,in order that the real properties of the browser are hidden with the static ones after noise adding in accordance with the realistic distribution based on Markov chain.In addition,open source fingerprinting scripts are used to verify the effectiveness of the method.Thirdly,by combining with detection and defense of browser fingerprinting,this thesis designs and implements a prototype system for mobile device privacy protection that supports intra-domain identification.The testing results demonstrate the validity of its support of intra-domain identification and prohibition of cross-domain linking,and tests the impact on browser performance.In summary,targeting at the problem of privacy leakage with browser fingerprinting based user tracking,this thesis takes an in-depth research and develops automatic detection technology,and thus realizes the automatic detection of mobile browser fingerprinting.Based on this,for browser fingerprint features with high rate of identification and usage,this thesis studies the defense of browser fingerprinting and develops a prototype system for mobile device privacy protection that supports intradomain tracking and prohibits cross-domain linking. |
