Research And Implementation Of Cryptocurrency Transaction Tracing Method Based On Traffic Analysis

Cryptocurrencies based on blockchain technology are developing rapidly,and the transaction scale of cryptocurrencies represented by Bitcoin continues to rise and hit new highs repeatedly.The decentralization,anonymity,and convenience of allowing users to send funds anywhere in the world provide opportunities for criminals to carry out illegal activities such as money laundering and black-market trading.The regulation of cryptocurrency transactions is imminent.In view of the difficulty of supervision and traceability brought about by the characteristics of anonymous addresses,decentralized transactions,and decentralized network of cryptocurrencies,we study the tracing mechanism of cryptocurrency transactions,propose a tracing method for cryptocurrency transactions based on traffic analysis,and combine transaction layer information to optimize the tracing effect,and validate the research protocol in real-world scenarios.The main achievements and innovations of the paper include:(1)We propose a semi-supervised learning-based method for nodeoriginated transaction identification.This method can identify the transaction created by the node itself from the transaction broadcast by the node,so as to associate the transaction with the IP address of the originating wallet node,thereby realizing the deanonymization of the transaction.This paper analyzes the feasibility and challenges of domestic regulatory authorities using ASes to collect node traffic,and proposes a more practical probe-based traffic collection method.We performe feature extraction and data set construction based on the collected traffic,and propose a nodeoriginated transaction identification algorithm based on semi-supervised learning for the first time.We mark the training set by proposing a pseudo-label marking algorithm,and use the trained model to identify the transactions created by the node itself,so as to realize the initial association between node and transactions created by nodes.This paper verifies and tests the algorithm in the Bitcoin network under the public network environment,and compares it with the existing tracing scheme based on machine learning algorithm.The identification algorithm has the advantages of high precision and small interference.(2)We propose a cross-layer collaborative analysis tracing method that combines transaction clustering results and node-originated transaction identification results.We propose a wallet-oriented originating transaction clustering method for the first time,which can cluster cryptocurrency transactions that may originate from the same wallet within a certain time window,and can help solve the problem that cryptocurrency transactions are scattered and messy and difficult to analyze effectively.The clustering results helps to improve the precision of transaction tracing method on the cryptocurrency network layer.By designing a cross-layer collaborative analysis tracing method that combines transaction clustering results and node-originated transaction identification results,the tracing effect is optimized and the precision of cryptocurrency transaction tracing is improved.Through experimental analysis and comparison,this scheme can effectively improve the precision and recall of node-originated transaction identification.(3)We study the application scheme of the transaction tracing method proposed in this paper in real-world scenarios.We design and develop a cryptocurrency cross-layer collaborative transaction tracing prototype system.The system supports basic functions such as personalized retrieval of blockchain transaction information and blockchain network node surveying and mapping.In addition,using the node-originated transaction identification model,the system can monitor node-originated transactions and identify suspicious originating transactions of nodes.In the multi-node scenario,we build a transactionoriented list of suspicious originating nodes,maintain the association mapping database,and support the query of transaction originating nodes.The system can effectively verify the research method of this paper in real-world scenarios and realize the tracing of cryptocurrency transactions.