ZKP-HGNN: A Study on Improving Zero-Knowledge Proof (ZKP) Based on Heterogeneous Graph Neural Networks for Anonymous Digital Identity Sharing in Blockchain

ZKP-HGNN: A Study on Improving Zero-Knowledge Proof (ZKP) Based on Heterogeneous Graph Neural Networks for Anonymous Digital Identity Sharing in Blockchain

연구 분야: Verification

학회: International Conference on Neural Information Processing

Blockchain technology has revolutionized digital transactions by offering decentralized, secure, and transparent ledger systems across diverse domains like finance, supply chain management, and healthcare. However, its inherent transparency poses challenges for privacy, especially concerning sensitive data. This paper explores privacy-preserving mechanisms in blockchain, focusing on Zero-Knowledge Proofs (ZKPs) and Graph Neural Networks (GNNs). ZKPs enable verifiable transactions without revealing underlying data, crucial for maintaining confidentiality in public blockchains. Variants like zk-SNARKs and zk-STARKs mitigate trust issues during setup and enhance computational efficiency. Meanwhile, GNNs leverage graph structures to model complex relationships, applicable in anomaly detection and secure communications within blockchain frameworks. This study innovatively combines ZKPs with Heterogeneous Graph Neural Networks (HGNNs) to enhance privacy in blockchain-based digital identity sharing. Methodologically, it proposes a ZKP-HGNN framework integrating identity representation, real-data authentication, and privacy-preserving transactions. Experimental validations on a private blockchain network demonstrate feasibility and efficiency. This research contributes novel insights and methodologies to fortify blockchain privacy, crucial for broader adoption in sensitive applications requiring data confidentiality.