Hidden Service Website Response Fingerprinting Attacks Based on Response Time Feature

摘要: It has been shown that website fingerprinting attacks are capable of destroying the anonymity communicator at traffic level. This enables local attackers to infer contents encrypted by using packet statistics. Previous researches on hidden service tend focus active attacks; therefore, reliability attack conditions and validity test results cannot be fully verified. Hence, it is necessary reexamine from perspective attacks. In this paper, we propose a novel Website Response Fingerprinting (WRFP) Attack based response time feature extremely randomized tree algorithm analyze information fingerprint. The objective monitor pages, types, mounted servers. WRFP relies dataset. addition simulated mirroring, two different mounting modes taken into account, same-source server multisource server. A total 300,000 page instances within 30,000 domain sites collected, comprehensively evaluate classification performance proposed WRFP. Our show TPR webpages remain greater than 93% in small-scale closed-world test, tolerating up 10% fluctuations time. also provides higher accuracy computational efficiency traditional classifiers challenging open-world test. indicates importance feature. suggest monitoring types improves judgment effect classifier subpages.