A High-Performance IPFS Service Architecture Compatible with Locator/Identifier Separation ICN
Tianhao Zhang, Jiali You, Li Yang · 2025
As the most popular decentralized service architecture, the availability of InterPlanetary File System(IPFS) for content fetch and function computation services is constrained by inefficient retrieval and rigid routing mechanisms. As a representative of new-generation networks, Information-centric networking (ICN) achieves fast content location and flexible routing capabilities, making ICN a crucial method to optimize IPFS. However, existing methods lack protocols and identifiers matching mechanisms between IPFS and ICN, making compatibility with the IP-based IPFS protocol stack and IPFS peers challenging. Therefore, we propose a new IPFS service architecture compatible with locator/identifier separation ICN, HPIPFS, which leverages network devices' growing computing and storage capabilities to integrate ICN edge nodes with IPFS peers at the application layer. HP-IPFS converts IPFS requests into ICN requests at ICN edge nodes, thereby bridging ICN resources with IP-based IPFS peers. Simultaneously, IPFS requests can utilize ICN's fast content location capabilities and flexible routing mechanisms to reduce content retrieval latency and improve the routing efficiency of computation requests. In different topology scales, experimental results demonstrate that compared to the original IPFS architecture, HP-IPFS achieved a maximum reduction of 34.26% in Time-To-First-Block and 34.31% in data block fetch latency. Meanwhile, HP-IPFS increased the computing task completion rate and computing resource utilization by 41.35% and 43.5%. At the same time, HP-IPFS demonstrates significant performance advantages in simulations at large realworld network scale.