Stealthy Clustering of Ethereum Transactions Using Time Leakage from Fixed Nodes

Abstract

In recent years, data mining with users as entities has become a fundamental tool for digital currency investigations. In Ethereum, researchers can perform transaction clustering based on users' behavioral habits and usage patterns in specific scenarios. However, the major shortcoming of these technologies is that they cover only a small fraction compared to the huge number of transactions. In this paper, we propose a stealthy clustering of Ethereum transactions using time leakage from fixed nodes (CIFO), which is applied to all transactions, while also being stealthy and difficult to detect. Leveraging timestamps relayed from $N$ fixed nodes, CIFO can better describe the network properties of transactions, in compared with the first $N$ timestamps. The experimental results show that CIFO can determine whether two transactions enter the network from the same source with the precision of 87%.