Malicious Transaction Deanonymity in Web 3.0

Web 3.0 is the next-generation Internet of value based on blockchain technology, bringing massive new innovations and opportunities in the economic, social, and financial fields. Web 3.0 places a strong emphasis on ensuring the privacy and control of personal data. Consequently, there has been a surge in the popularity of mixing services, which enhance transaction anonymity. Unfortunately, the robust anonymity provided by these services has also attracted criminals who engage in money laundering and fraudulent activities. These security concerns pose a significant threat to the Web 3.0 ecosystem and financial stability. To ensure a secure financial environment and deanonymize malicious transactions in Web 3.0, this chapter proposes a framework based on GNN called MixBroker. Its purpose is to disrupt Tornado Cash’s anonymity, the most popular mixing service on the Ethereum blockchain, by establishing connections between mixing addresses through node-pair link prediction. To facilitate further analysis, we utilize mixing transaction data on Ethereum to build a Mixing Interaction Graph (MIG). To capture the unique characteristics of mixing addresses, features are extracted from multiple perspectives within the MIG. In parallel, we develop a link prediction module as the core component of MixBroker. It leverages the interconnected structure of the MIG to calculate the likelihood of correlation between mixing addresses. To address the issue of insufficient ground-truth data, we amass a substantial volume of actual mixing transactions sourced from Ethereum and Tornado Cash. Additionally, we incorporate the principles of the Ethereum Name Service to establish a dependable ground truth.