Sensitivity analysis for a Bitcoin simulation model

Abstract

Bitcoin is a popular and widely traded cryptocurrency. The Bitcoin blockchain technology makes it easy for users to conduct pseudo-anonymous financial transactions. However, it also facilitates criminals to secrete their actual identities from law enforcement agencies. Heuristic-based address clustering is the subject regarding Bitcoin de-anonymization. But no heuristic algorithm has a known or potential error rate due to the lack of ground truth. This paper uses sensitivity analysis to validate and verify a constructed Bitcoin simulation model. The evaluation and validation processes examine the model behavior and model outputs from multiple simulation runs to demonstrate fidelity and credibility. The analysis results show no model uncertainties, and the simulation model is stable and can effectively simulate Bitcoin transactions. With a reasonable number of nodes and transaction volumes in the simulated network, the simulation model can be used to verify the effectiveness of two widely used heuristic-based address clustering algorithms and measure the corresponding error rates.