Design and analysis of anonymous communications for emerging applications
Date
Authors
Major Professor
Advisor
Committee Member
Journal Title
Journal ISSN
Volume Title
Publisher
Altmetrics
Authors
Research Projects
Organizational Units
Journal Issue
Is Version Of
Versions
Series
Department
Abstract
Emerging distributed applications like distributed storage, P2P-based distributed applications and grid-based applications have different requirements and privacy challenges which cannot be solved efficiently using traditional anonymity protocols. For example, rerouting-based techniques are vulnerable to intersection attacks in P2P-based anonymity systems (Tarzan, Morphmix
etc.) due to the inherent churn, multi-hop forwarding protocols have high overhead in data
grid applications, and distributed hash table based platforms are vulnerable to privacy leaks due to mapping between nodes and data.
We have tried to address some of these issues in this research. We have proposed an incentive-based forwarding protocol for P2P anonymity systems, which uses game theory to develop peer strategies in participating and forwarding traffic for other peers. Using simulations, we show that the protocol performs
appreciably well for realistic scenarios of churn and varying degrees of malicious node presence. We also outline a possible implementation of
the scheme which enhances anonymity through peer participation and thereby decreasing the
possibility of intersection attacks. We have also proposed a 2-hop forwarding protocol for data-grids which uses the inherent trust of grid environments to select forwarding peers. Through simulations of realistic grid scenarios, we show that an appreciable degree of anonymity can be achieved without incurring huge overhead.
In the second part of the thesis, we address privacy leaks from distributed hash tables. We have developed an analytical model, based on the
information-theoretic metric of entropy, which compares different DHT designs. Our model shows that ring-based structure like Chord is the least vulnerable to information leak. We also compare traditional DHT designs to unstructured networks like Gnutella. We believe such a model can be useful in future DHT designs whereby privacy is an important concern.