Interactive Oracle Proofs of Proximity to Codes on Graphs
Hugo Delavenne, Tanguy Medevielle, Élina Roussel
TL;DR
This work introduces an Interactive Oracle Proof of Proximity (IOPP) tailored for codes on graphs, inspired by the FRI protocol, to test proximity to $\mathcal{C}[\Gamma,k]$ with oracle access and strong soundness. The core method, the Flowering protocol, folds the graph structure down to a single-vertex flower while preserving local views, enabling proximity testing against a Reed-Solomon base code $\mathsf{RS}[n,k]$ with improved soundness and relaxed field-size requirements $|\bbF|>\log N$. Key contributions include a formal dimension lower bound for graph-based codes, a commit-soundness analysis, and a constructive Cayley-multigraph instantiation that achieves positive rate with $o(1)$ minimum distance, along with concrete parameter trade-offs against the FRI protocol. The results offer a practical pathway to code-based SNARKs with flexible alphabets and scalable graphs, potentially enabling efficient arithmetization in frameworks like PlonK or R1CS. Future work aims to broaden the flowering construction beyond two-way cuts and explore richer equivalence classes of cut-graphs to further enhance applicability.
Abstract
We design an Interactive Oracle Proof of Proximity (IOPP) for codes on graphs inspired by the FRI protocol. The soundness is significantly improved compared to the FRI, the complexity parameters are comparable, and there are no restrictions on the field used, enabling to consider new codes to design code-based SNARKs.
