Eclipse attacks on Monero peer-to-peer network
summary
Collected TCP flow peerlist data from three global monitors (US, Europe, Singapore) over three weeks and validate a connection-inference algorithm that uses per-peer frequency clustering and adaptive thresholds to filter interactions, achieving ~80% inference accuracy that improves with longer observation. The resulting map labels 4,800 peers mapped and highlights 14 highly connected supernodes whose overlap makes simulated Eclipse attacks cheaper but still requires removing roughly 50 nodes to collapse the largest connected component.
AI-written summary · about
chapters (15)
- 0:00Decentralization in peer-to-peer networks
- 2:37Monero P2P network landscape
- 6:11Monero peer protocol changes
- 9:24Mapping Monero network
- 13:39Monero P2P topology
- 16:53Simulating eclipse attacks
- 19:49Eclipse attack consequences
- 20:32Decentralization vs security trade-offs
- 21:59Q&A on ZK proofs
- 22:45Cost of eclipse attacks
- 24:16Super nodes control
- 25:21Q&A on super nodes
- 27:10Geographical super node distribution
- 27:52Q&A on network topology
- 28:36Summary
related talks
43mWho are the people in your neighborhood? Tracking 'Shady' CIDR blocks
59mCatching Malware En Masse: DNS and IP Style
17mPrivateEye: Scalable and Privacy Preserving Compromise Detection in the Cloud
32mActive Fault Finding in Networks
05h 42mbitcoin++ privacy edition | Riga, Latvia | Day 1
37mHiding the Rings: The Importance of P2P Encryption in Monero
44mNetsec is Dead(?): Modern Network Fingerprinting for...
17mLIVE w/ Josh S (Airfoil Capital) & Lee Clagett (The Monero Project) at MoneroKon 2019! EPI #71