TMAS: A Transaction Misbehavior Analysis Scheme for Blockchain

Abstract

The emergence of blockchain-based cryptocurrencies, such as Bitcoins, has presented a promising alternative for e-payment methods, owing to their unique features of decentralization and anonymity. The usage of these currencies has grown exponentially, particularly in anonymous e-payment and without any trusted third party. However, the decentralized and anonymous nature of these currencies has also resulted in misbehaviors, e.g., money laundering. Therefore, detecting transaction misbehaviors has garnered increasing attention. In this paper, we propose TMAS, a transaction misbehavior analysis scheme for blockchain-based cryptocurrencies. We propose various transaction analysis approaches, feature extraction algorithms, and detection models for misbehaviors, including money laundering. We have implemented a real experimental system to detect misbehaviors, including money laundering, in blockchain-based cryptocurrencies such as Bitcoins. The proposed system includes ten features in the transaction graph, two heuristic money laundering models, and an analysis method for account linkage, which identifies accounts that are distinct but controlled by an identical entity. To verify the effectiveness of our proposed indicators and models, we have analyzed a sample of 100M transactions and computed transaction features, leading to the identification of some suspicious accounts. Moreover, the proposed methods can be applied to other cryptocurrencies, no matter token-based such as Bitcoins or account-based such as Ethereum.