Analyzing and Enhancing Routing Protocols for Friend-to-Friend Overlays

Datum

25.07.2016

Zeit

13:00 - 14:00

Sprecher

Dipl.-Math. Stefanie Roos

Zugehörigkeit

Institut für Systemarchitektur, Datenschutz und Datensicherheit

Sprache

en

Hauptthema

Informatik

Andere Themen

Informatik

Beschreibung

The threat of surveillance by governmental and industrial parties is more eminent than ever. As communication moves into the digital domain, the advances in automatic assessment and interpretation of enormous amounts of data enable tracking of millions of people, recording and monitoring their private life with an unprecedented accurateness. The knowledge of such an all-encompassing loss of privacy affects the behavior of individuals, inducing various degrees of (self-)censorship and anxiety. Furthermore, the monopoly of a few large-scale organizations on digital communication enables global censorship and manipulation of public opinion. Thus, the current situation undermines the freedom of speech to a detrimental degree and threatens the foundations of modern society. Anonymous and censorship-resistant communication systems are hence of utmost importance to circumvent constant surveillance. However, existing systems are highly vulnerable to infiltration and sabotage. In particular, emph{Sybil attacks}, i.e., powerful parties inserting a large number of fake identities into the system, enable malicious parties to observe and possibly manipulate a large fraction of the communication within the system. Friend-to-friend (F2F) overlays, which restrict direct communication to parties sharing a real-world trust relationship, are a promising countermeasure to Sybil attacks, since the requirement of establishing real-world trust increases the cost of infiltration drastically. Yet, existing F2F overlays suffer from a low performance, are vulnerable to denial-of-service attacks, or fail to provide anonymity. Our first contribution in this thesis is concerned with an in-depth analysis of the concepts underlying the design of state-of-the-art F2F overlays. In the course of this analysis, we first extend the existing evaluation methods considerably, hence providing tools for both our and future research in the area of F2F overlays and distributed systems in general. Based on the novel methodology, we prove that existing approaches are inherently unable to offer acceptable delays without either requiring exhaustive maintenance costs or enabling denial-of-service attacks and de-anonymization. Consequentially, our second contribution lies in the design and evaluation of a novel concept for F2F overlays based on insights of the prior in-depth analysis. Our previous analysis has revealed that greedy embeddings allow highly efficient communication in arbitrary connectivity-restricted overlays by addressing participants through coordinates and adapting these coordinates to the overlay structure. However, greedy embeddings in their original form reveal the identity of the communicating parties and fail to provide the necessary resilience in the presence of dynamic and possibly malicious users. Therefore, we present a privacy-preserving communication protocol for greedy embeddings based on anonymous return addresses rather than identifying node coordinates. Furthermore, we enhance the communication's robustness and attack-resistance by using multiple parallel embeddings and alternative algorithms for message delivery. We show that our approach achieves a low communication complexity. By replacing the coordinates with anonymous addresses, we furthermore provably achieve anonymity in the form of plausible deniability against an internal local adversary. %, consisting of the complexity of the message delivery in addition to the complexity of maintaining uptodate coordinates. Complementary, our simulation study on real-world data indicates that our approach is highly efficient and effectively mitigates the impact of failures as well as powerful denial-of-service attacks. Our fundamental results open new possibilities for anonymous and censorship-resistant applications.

Letztmalig verändert: 25.07.2016, 10:02:53