Privacy-Preserving Cross-Chain Asset Tracking: A Threshold PSI Protocol with Real-World Exploit Validation

Abstract

Cross-border financial investigations often face difficulties due to isolated data systems, especially when different Financial Intelligence Units (FIUs) independently track illegal funds moving across different blockchain networks. This paper presents an enhanced multi-chain $t$-out-of- $n$ threshold Private Set Intersection (PSI) protocol for high-throughput, privacypreserving jurisdictional deconfliction. Built upon Shamir's Secret Sharing over the $\mathbb{Z}_{2^{127}}$ Mersenne prime field, the proto-col introduces a high-performance cryptographic field-mapping mechanism leveraging the BLAKE3-XOF. The protocol is formally validated under the semi-honest adversarial model, with three theorems establishing sub-threshold privacy, intersection completeness, and a negligible false-positive rate of 2−127. We evaluate scalability on three heterogeneous blockchain corpora (Bitcoin UTXO and Ethereum account-based models) scaled up to 500,000 records per ledger. The BLAKE3-based mapping achieves a 1.88× throughput improvement over a SHA-256/512 baseline, reaching up to $1.55 \times 10^{6}$ operations per second. Experimental validation across a multi-ledger testbed using the Elliptic++, BABD-13, and XBlock-ETH datasets in a three-jurisdiction FIU deconfliction setting confirms correct threshold enforcement, full recovery of all three-way illicit intersections without false positives, and near-linear runtime behaviour. In a larger fivejurisdiction setting, the system reliably identifies all threshold intersections while maintaining zero reconstruction errors. The proposed protocol is relevant for FIUs, blockchain analytics researchers, and cybersecurity practitioners engaged in crossborder illicit asset tracing and privacy-preserving data collaboration.