An automated method for detecting suspicious nodes in Bitcoin address graph

Abstract

Financial innovation has entered a new era with cryptocurrencies. Bitcoin is the first decentralised cryptocurrency and the most popular in the world. The main features of this new technology are immutability, decentralised trust and anonymity. Bitcoin anonymous and untraceable system facilitates cash-out and laundering of cybercrime proceeds. Bitcoin currency flow provides an address graph that assigns the flow of bitcoin between two addresses. Identifying suspicious nodes in the bitcoin network is similar to the problem of recognising the origins in the contact network of different applications, i.e., virus propagation, rumour source in social networks, and poison spread in water networks. In order to investigate money laundering in bitcoin, we proposed an automated method to identify suspicious addresses in the bitcoin address graph. We chose two centrality measures to be calculated on the graph. The measures are betweenness centrality and closeness centrality. The nodes with the highest values for the measurements are suspicious. The accuracy of the proposed method is further investigated by comparing the fraudulent candidate nodes with other scenarios. It is shown that the identified nodes are correct candidates for further investigations.