Paper 2024/1544

PoUDR: Proof of Unified Data Retrieval in Decentralized Storage Networks

Abstract

Decentralized storage networks, including IPFS and Filecoin, have created a marketplace where individuals exchange storage space for profit. These networks employ protocols that reliably ensure data storage providers accurately store data without alterations, safeguarding the interests of storage purchasers. However, these protocols lack an effective and equitable payment mechanism for data retrieval, particularly when multiple data queriers are involved. This necessitates a protocol that ensures both data integrity and fair compensation for data providers. In decentralized storage, data is fragmented into small blocks and stored across multiple nodes, a process known as data swarming. Due to this property, traditional data exchange protocols are inadequate in terms of communication and economic efficiency. We propose the Proof of Unified Data Retrieval protocol (PoUDR). PoUDR incorporates ZK-SNARK to facilitate a fair data exchange protocol. PoUDR reduces the number of blockchain transactions for both single block and data swarming retrieval. The protocol requires only a single key-revealing transaction submitted by the provider to the blockchain for each data block. This architecture allows for further optimization of transaction numbers through a batched proof technique on the provider's side. This approach necessitates only $N_P$ transactions within a specific time frame when data consisting of $N_D$ blocks, provided by $N_P$ providers, is queried by $N_Q$ queriers. This work provides a comprehensive definition for Secure Swarming Data Exchange (SSDE), including security assumptions. Also it offers a detailed game-based security analysis for the PoUDR protocol. Moreover, the PoUDR protocol has been fully integrated into the Bitswap protocol (IPFS). Within this integration, our proposed Relaxed Groth16 algorithm addresses the significant technical challenge of generating zero-knowledge proofs, leading to substantial cost reductions for overall feasibility of secure data retrieval in decentralized storage networks.