| Cryptocurrencies have brought financial privacy protection to users,but also provided financial tools for some criminal activities.There is a large amount of illegal funds in the cryptocurrency system,and the incomplete blacklist system of important nodes in the cryptocurrency system,such as digital currency exchanges,allows illegal funds to flow freely.In addition,the existence of tools such as coin mixers makes it more convenient for illegal funds to escape,and the current detection methods for coin mixing transactions lack credibility due to a lack of real samples as controls.The inadequacy of detection methods further leads to the inability to trace illegal funds in coin mixing transactions.To address the shortcomings of identifying and tracking illegal fund transactions in the cryptocurrency system,we propose a blacklist system that allows on-chain users to vote and report illegal funds that have not yet been transferred,and propose detection methods for coin mixing transactions and identification methods for intermediate addresses in coin mixing to trace illegal funds that have already been transferred.The specific contributions are as follows:(1)Combining smart contract and editable blockchain technologies,a public chain-based blacklist solution based on a voting mechanism is proposed.First,the process of adding a fund to the blacklist is designed from the perspective of game theory using smart contract technology,using reward and punishment measures to encourage honest strategies from the players.For confirmed illegal funds,a field called "fund anomaly degree" is designed to reduce the probability of a transaction being selected when it is added to the blockchain,replacing the traditional way of executing a blacklist system.Experimental results show that in the overall block packaging process,the average probability of a transaction with the highest anomaly degree being selected is only19.5%,and the average probability of a transaction that has already been added to the blockchain being selected is reduced to 68.4%.(2)Combining the characteristics of the Coin Join mixing model and the numerical features of transaction samples,a detection method for mixed transactions based on Coin Join is proposed.First,a general detection method for Coin Join mixed transactions is obtained based on the concept of Coin Join and its limitations on the size of the anonymous transaction set and the mixed amount.Then,it is combined with some features of the Coin Join implementation platform Wasabi to obtain the basic detection method for Wasabi.Next,a reliable verification dataset is obtained from the Wasabi service interface,and the rule parameters in the detection method are adjusted to obtain an improved detection method for Wasabi.Experimental results show that after analysis and feedback from the verified dataset,the recall rate of the improved detection method reaches 100% and the precision rate is above 99%.(3)Through machine learning,a method for identifying mixing intermediate addresses and tracking illegal funds in mixing transactions based on the transaction behavior patterns of the addresses is proposed.Firstly,starting from the intermediate addresses in mixing transactions,five transaction behavior patterns are identified for the addresses holding mixed funds,and their correlation with illegal funds is analyzed to propose a method for tracking illegal funds after mixing.Next,the features of mixing intermediate addresses are analyzed and summarized.A random forest classifier is used to rank the importance of features for a transaction address,and the top 6important features are selected to form a feature vector.A decision tree algorithm is used to construct a classification model for identifying mixing intermediate addresses.Experimental results show that the decision tree classification model achieves a precision rate of 95.0% and a recall rate of 97.3% for identifying mixing intermediate addresses. |
PDF Full Text Request