TY - THES A1 - Mitseva, Asya T1 - State of affairs in online user privacy : assessing real-world user information leakages and designing effective and practical countermeasures T1 - Aktueller Stand der Privatsphäre von Internetnutzern : Bewertung realer Datenlecks und Entwicklung wirksamer und praktischer Gegenmaßnahmen N2 - The Tor network ensures confidentiality of user data over the Internet and hides the IP addresses and relationships between communication partners to avoid user profiling and censorship. However, Tor cannot hide the number, direction, and timing of transmitted packets. Two types of attackers – autonomous systems (AS) and malicious Tor nodes – can exploit this for sophisticated attacks, such as website fingerprinting (WFP) and end-to-end traffic correlation, or fingerprinting attacks revealing the Tor users’ location. On the other hand, in case of a malicious first Tor node or a malicious AS on the path between the user and the Tor network, the scalability of prior WFP attacks remains questionable due to impractical assumptions. This dissertation aims to provide a more realistic assessment of the exposure of Tor users to passive deanonymization attacks from different perspectives. First, we conduct a comprehensive review of fundamental Border Gateway Protocol (BGP) threats. We present a methodology for evaluating existing BGP security proposals and provide an extensive and up-to-date survey of proposals intended to make BGP secure and methods for detecting and mitigating routing instabilities. Based on the identified research gaps and unresolved challenges, we analyze the current level of resilience of Tor and two alternative Tor path selection methods against malicious ASs. While we detect a notable increase in the Tor’s resilience to BGP attacks over the years, when measured from a Tor user to its first Tor node, the resilience provided by the alternative path selection methods on the way back, i.e., from the first Tor node to a Tor user, has decreased notably compared to the resilience measured on the forward path for these methods. Moreover, these methods tend to select first Tor nodes in a predictable manner, resulting in the exposure of users’ locations to Tor middle nodes. We also show novel attacks against a state-of-the-art method for measuring bandwidth capabilities of Tor nodes, allowing a low-bandwidth attacker to gain notably higher bandwidth estimates than theoretically defined. We propose a novel lightweight defense against malicious Tor nodes, performing WFP or end-to-end traffic correlation, that splits user traffic over multiple Tor paths. Our defense is fully implemented on the user side, reducing deployment costs and making it independent of the underlying anonymization network. We also propose a metric that is capable of quantifying the user’s location leakage considering a malicious Tor middle node. This metric enables Tor users to avoid specific first Tor nodes by strategically selecting those from which their location cannot be revealed. Finally, we examine the real-world scalability of WFP, especially against Tor users who visit several pages of a single website consecutively. We propose both a novel WFP attack and efficient strategies for adapting existing methods to account for sequential visits of pages within a website. KW - Anonyme Kommunikation KW - Border Gateway Protocol (BGP KW - Datenschutz im Internet KW - Datenverkehrsanalyse KW - Onion routing; Tor network KW - Traffic analysis KW - Website fingerprinting KW - Website fingerprinting defenses Y1 - 2025 U6 - https://doi.org/10.26127/BTUOpen-7214 PB - Brandenburgische Technische Universität CY - Cottbus-Senftenberg ER -