Dandelion: Redesigning the Bitcoin Network for Anonymity

Shaileshh Bojja Venkatakrishnan came to MIT to give a talk on his work redesigning the Bitcoin network. I’ll provide an overview here, but for more details, I refer you to his paper.

It is without a doubt that Bitcoin has become more and more popular. The price of Bitcoin at the time of this post is ~1350 dollars. One reason that it is so popular is that users believe it is “untraceable” and “anonymous,” but numerous attacks have shown this not to be true. The focus of this work is to prevent deanonymization through network based attacks.

He proposes a redesign of the Bitcoin network, which is simple to implement and has near optimal anonymity guarantee. The adversary model is the following:

  • fraction of p nodes are spies
  • identities unknown
  • observe all metadata
  • honest-but-curious
Metric for anonymity

The goal of Dandelion is to design a distributed spreading protocol that minimizes the maximum precision and recall achievable by the adversary.

The algorithm has two phases: an anonymity phase and spreading phase. I won’t go through the algorithms here, but the key insight is that it uses information theoretic mixing properties. There are two graphs, an anonymity graph and a spreading graph. Dandelion changes the anonymity graph frequently

He shows in his results that Dandelion shows promise as a lightweight solution for certain classes of Bitcoin network adversaries. This is an interesting attempt to try to redesign the Bitcoin protocol. His paper has way more of the “meaty” details of the protocol.