Reducing Privacy of CoinJoin Transactions: Quantitative Bitcoin Network Analysis

Privacy within the Bitcoin ecosystem has been critical for the operation and propagation of the system since its very first release. While various entities have sought to deanonymize and reveal user identities, the default semi-anonymous approach to privacy was judged as insufficient and the community developed a number of advanced privacy-preservation mechanisms. In this study, we propose an improved variant of the multiple-input clustering approach that incorporates advanced privacy-enhancing techniques. We examine the CoinJoin-adjusted user graph of Bitcoin through quantitative network analysis and draw conclusions on the effectiveness of our proposed clustering method compared to naive multiple-input clustering. Our findings indicate that CoinJoin transactions can significantly distort commonly applied address clustering approaches. Moreover, we demonstrate that Bitcoin's user graph has become less dense in recent years, concurrent with the collapse of several independent user clusters. Our results contribute to a more comprehensive understanding of privacy aspects in the Bitcoin transaction network and lay the groundwork for developing enhanced measures to prevent money laundering and terrorism financing.