Bitcoin Coin Mixing Transaction Detection And Pattern Analysis

As cryptocurrencies like Bitcoin become more popular and de-anonymization technologies advance,users are increasingly concerned about protecting their personal privacy.In response,coin mixing services have emerged to enhance the privacy of Bitcoin transactions.However,misusing these services can potentially increase illegal activities,which poses a risk to society.The coin mixing process in Bitcoin is a black box for most people,and this thesis aims to uncover the characteristics and patterns of such services through a detailed analysis of the transaction data.The research hopes to provide valuable insights for Bitcoin users and regulators,and to help seek a balance between protecting users’ privacy and preventing illegal activities from breeding.This thesis focuses on Bitcoin coin mixing services.Coin mixing services can be divided into two main categories: distributed coin mixing services and centralized coin mixing services.The details of the research are as follows:(1)Study on identifying distributed coin mixing services.These services mainly use the Coin Join protocol,which merges transactions from multiple users to obscure their transactions.This research focuses on three wallets that support Coin Join,and they all obfuscate the correspondence between the two sides of a transaction by generating anonymous sets.By analyzing and improving existing heuristics,the research proposes three new heuristics for accurately identifying the different types of coin mixing transactions.Additionally,13 transaction features are used alongside four machine learning models to classify transaction dataset and identify distributed coin mixing services.Experiments show that the random forest model performs best with an accuracy of 99.98%.(2)Analysis of the peeling chain model in centralized coin mixing services.The peeling chain model is a common method used in centralized coin mixing services.This research examines coin mixing transactions in the Bitcoin Fog service and provides a detailed definition of the three parts of the peeling chain: user input transactions,merge transactions,and distribution transactions.By extracting coin mixing transaction datasets from the Wallet Explorer and analyzing these three components,the research explores the transaction process and uncovers patterns and characteristics.Subsequent distribution of transaction search algorithm is proposed to identify distribution transactions in the peeling chain.The experiment compares the peel chain data detected under different initial distribution values and verifies it through specific transactions,confirming the effectiveness of this coin mixing transaction recognition method.