ChurnAddr Strategy in the Bitcoin Network: Assessing the Applicability of using Churn and Address Propagation Exploits in the Bitcoin Network to Perform Adversary Actions

Larsen, Rune Willum ; Frie, Alex Grøndahl

4. term

Software, Master

2018

2018-06-08

45

This work extends a prior study on whether Bitcoin’s address propagation and churn can be exploited to create a network topology favorable to an adversary. We propose the ChurnAddr strategy, in which an adversary runs highly stable nodes with full Bitcoin Core features and repeatedly broadcasts ADDR messages (up to 10 IP addresses) to increase inbound connections. Because long-term measurements on the live network are impractical, we build a data-driven, event-based simulator with fine-grained time, dynamic topology, deterministic randomness, and optimized performance, informed by code review and reverse engineering of Bitcoin Core as well as churn and latency data collection. Across multiple scenarios, we find that churn alone can raise an adversary’s inbound connections and that block propagation can be influenced, including delayed for selected participants, potentially conferring a block race advantage to a miner. We do not find strong correlations for deriving node-to-node latency, but provide new insight into operating systems in the network. Overall, we assess ChurnAddr’s applicability: it is unsuitable for eclipse attacks, but can disrupt block propagation. We also contribute foundational methods and an improved simulation framework for long-term studies of the Bitcoin network.

Dette arbejde viderefører en tidligere undersøgelse af, om Bitcoins adressepropagation og churn kan udnyttes til at skabe en netværkstopologi, der begunstiger en modstander. Vi foreslår ChurnAddr-strategien, hvor en adversary etablerer og vedligeholder meget stabile noder med fulde Bitcoin Core-funktioner og jævnligt udsender ADDR-beskeder (op til 10 IP-adresser) for at øge antallet af indgående forbindelser. Da langtidsmålinger på det virkelige netværk er upraktiske, udvikler vi en data-drevet, hændelsesbaseret simulator med fin tidsopløsning, dynamisk topologi, deterministisk tilfældighed og optimeret ydeevne, understøttet af code review og reverse engineering af Bitcoin Core samt indsamling af churn- og latensdata. I flere scenarier viser vi, at churn i sig selv kan øge en adversarys indgående forbindelser, og at blokpropagering kan påvirkes, herunder forsinkes for udvalgte deltagere, hvilket kan give en fordel i blokløbet for eksempelvis en miner. Vi fandt ikke stærk korrelation til at udlede node-til-node-latens, men bidrager med indblik i operativsystemer i netværket. Samlet set vurderer vi ChurnAddr-strategiens anvendelighed: den er ikke egnet som del af et eclipse-angreb, men kan bruges til at forstyrre blokpropagering. Derudover bidrager vi med grundlæggende metoder og et forbedret simulatorgrundlag for langtidsstudier af Bitcoin-netværket.

[This apstract has been generated with the help of AI directly from the project full text]

churn ; blockchain ; bitcion ; model ; simulator

Download PDF
View record in AAU Student Projects