A holistic approach towards peer-to-peer security and why proof of work won’t do

Abstract

Separation of identity and location is one of the key properties of peer-to-peer networks. However, this separation can be abused to mount attacks against the network itself. Our contribution in this matter is twofold: First, we present a security-first design for P2P networking based on self-certifying identifiers. It provides message authenticity, integrity of routing tables, and authenticated communication, is resistant (and not only resilient) against many typical peer-to-peer-specific attacks, and guarantees uniform identifier distribution. The second aspect of our contribution disproves the often-quoted assumption that proof-of-work-based identifier generation can sufficiently hinder certain peer-to-peer attacks such as the Sybil attack. This finding seriously questions previously proposed proof-of-work-based defence mechanisms and leads to the only conclusion possible: Proof-of-work-based measures to limit arbitrary identifier generation do not stand the test of reality.