The Dynamics Of Salsa: A Structured Approach To Large-scale Anonymity

Abstract

Anonymity provides a technical solution for protecting one's online privacy. Highly distributed peer-to-peer (P2P) anonymous systems should have better distribution of trust and more scalability than centralized approaches, but existing systems are vulnerable to attacks as they require nodes to have global knowledge of the system. To overcome these problems and to provide more secure, distributed organization for P2P anonymity systems, prior work proposed the Salsa system. Salsa is designed to select nodes to be used in anonymous circuits randomly from the full set of nodes, even though each node has knowledge of only a small subset of the network. It uses randomness, redundancy and bounds checking while performing lookups to prevent malicious nodes from returning false information without detection. In this thesis, we further investigate the dynamics of the Salsa system and propose to handle important system-level functionalities without giving advantages to attackers. We propose algorithms for joining and leaving of a node, splitting a group when its size reaches a maximum threshold, merging of two groups when a group's size reaches a minimum threshold and updating global contacts of nodes locally. We have introduced more randomness in the lookup procedure and made bounds checking more flexible. Finally, we implemented these dynamic events and developed a complete continuous time simulator for Salsa. Using this simulator, we present simulation results that show that still Salsa continues to have good lookup success in a dynamic environment with modest overheads in the system. These results also demonstrate the stability of Salsa in the presence of many peers joining and leaving.