Efficient Fine-grained Accountable Weighted Threshold Signature for Conditional Privacy-preserving Lending Transactions in Blockchain

Blockchain technology, with its characteristics of decentralization, immutability and traceability, provides a new platform and technical means for financial transactions, such as lending systems. There are critical security and efficiency demands: Firstly, due to the high volume of real-time concurrent lending transactions, the system must be highly efficient. Furthermore, to achieve fine-grained accountability for illegal lending approval under regulatory oversight, a weighted threshold architecture is required to enable multi-auditor reviews aligned with hierarchical differences among auditors (e.g., seniority or authority levels), and ensure compliance and fairness. Unfortunately, the existing work of threshold signatures face significant limitations: fully anonymous schemes lead to accountability challenges (e.g., difficulty in tracing malicious actors); node interactions during initialization result in prohibitively high overhead; excessive disclosure of identity information severely compromises signer privacy. These shortcomings render traditional schemes unsuitable for lending transactions in blockchain. In order to address these issues, we firstly propose an efficient fine-grained accountable weighted threshold signature (FAWTS). It enables non-interactive initialization without trusted third parties, achieves fine-grained accountability and introduces dynamic weighted thresholds. Then based on FAWTS, we design a conditional privacy-preserving lending transactions in blockchain, which assigns dynamic weighted thresholds for multiple auditors/lending transactions, enables fine-grained tracing of malicious auditors and achieves non-interactive auditor selection to meet high real-time processing demands. Finally, we formally prove that our proposed schemes achieve both the unforgeability and privacy requirements, while significantly reducing computational and communication overhead under extended functionalities.