A Dataset-Driven Validation of Structural Constraints in Network Traffic–Based Detection of Financial Crimes

Modern financial ecosystems are increasingly exposed to money laundering, fraud, phishing, and transaction obfuscation within encrypted and decentralized network environments. While prior research identifies detection challenges conceptually, limited work empirically validates how structural constraints manifest across real-world network datasets. This study presents a dataset-driven validation of structural constraints in network traffic–based detection of financial crime using three progressively scaled real-time datasets: NetTran3, NetTran4, and NetTran5. Quantitative analysis shows that encrypted traffic obscures 22.03% of transactions in NetTran3, rises to 39.29% in NetTran4, and declines to 18.24% in NetTran5, reflecting evolving encryption dynamics. Tumbling services—used for transaction obfuscation—are observed in 34.42%, 34.01%, and 40.22% of transactions, respectively, indicating increasing transactional fragmentation. Additional constraints include scalability pressure, cross-chain complexity, adversarial manipulation, and data imbalance. Rather than benchmarking detection algorithms, this study systematically evaluates how these infrastructural characteristics intensify detection difficulty as dataset size and complexity increase. The findings provide empirical grounding for literature-identified constraints and establish a structured foundation for advancing robust, scalable, and privacy-aware financial crime detection systems.