Behavioural Analysis for Money Laundering Activity in the Bitcoin Network

Abstract

Blockchain networks securely record transactions and enable decentralised transactions using cryptocurrencies. However, the pseudonymity nature of the participants makes the blockchain network a platform for illegal activities, such as money laundering, which poses significant threats to financial security and regulatory compliance. Money laundering activities undermine the integrity of financial systems, foster criminal enterprises, and enable tax evasion. This research explores the impact of timestamp-based (Time step) features in detecting money laundering activities within the Bitcoin network, utilising the Elliptic++ dataset. A correlation-based analysis revealed that the first block appeared in feature was the most strongly correlated with the Time step. Additionally, classification results highlighted XGBoost as the most effective classifier, with the first block appeared in feature identified as the most influential, based on Shapley values from the eXplainable Artificial Intelligence (XAI) technique.