SybilConf: Computational Puzzles for Confining Sybil Attacks

Abstract

In decentralized dynamic networks such as peer-to-peer networks, it is difficult to authorize identities without a centralized authority, since a node interacts with nodes previously unknown to itself. To optimize the quality of experience and exclude malicious nodes from such networks, various trust and reputation systems (TRS) have been proposed. However, most of these TRSs rely on properties such as trust attributed to certain persistent identities and the bootstrapping is difficult without a central trusted instance. Hence, TRSs are vulnerable to sybil attacks [1], where a number of artificial identities can be used by some malicious entity to compromise probabilistic protection mechanisms. These mechanisms usually rely on the assumption that most nodes in a network behave well. Unfortunately, this is not the case if an attacker maintains numerous identities. We present SybilConf, a simple scheme to increase costs for maintaining an identity, thus reducing the sybil attack's impact. The basic idea is to combine public/ private keying with a novel puzzle creation algorithm to enable a flexible, strong and identity-bound sybil protection.