International Association
for Cryptologic Research

A cryptographic accumulator is a succinct set commitment scheme with efficient (non-)membership proofs that typically supports updates (additions and deletions) on the accumulated set. When elements are added to or deleted from the set, an update message is issued. The collection of all the update messages essentially leaks the underlying accumulated set which in certain applications is not desirable. In this work, we define oblivious accumulators, a set commitment with concise membership proofs that hides the elements and the set size from every entity: an outsider, a verifier or other element holders. We formalize this notion of privacy via two properties: element hiding and add-delete indistinguishability. We also define almost-oblivious accumulators, that only achieve a weaker notion of privacy called add-delete unlinkability. Such accumulators hide the elements but not the set size. We consider the trapdoorless, decentralized setting where different users can add and delete elements from the accumulator and compute membership proofs. We then give a generic construction of an oblivious accumulator based on key-value commitments (KVC). We also show a generic way to construct KVCs from an accumulator and a vector commitment scheme. Finally, we give lower bounds on the communication (size of update messages) required for oblivious accumulators and almost-oblivious accumulators.

Proof-of-Replication (PoRep) plays a pivotal role in decentralized storage networks, serving as a mechanism to verify that provers consistently store retrievable copies of specific data. While PoRep’s utility is unquestionable, its implementation in large-scale systems, such as Filecoin, has been hindered by scalability challenges. Most existing PoRep schemes, such as Fisch’s (Eurocrypt 2019), face an escalating number of challenges and growing computational overhead as the number of stored files increases. This paper introduces a novel PoRep scheme distinctively tailored for expansive decentralized storage networks. At its core, our approach hinges on polynomial evaluation, diverging from the probabilistic checking prevalent in prior works. Remarkably, our design requires only a single challenge, irrespective of the number of files, ensuring both prover’s and verifier’s run-times remain manageable even as file counts soar. Our approach introduces a paradigm shift in PoRep designs, offering a blueprint for highly scalable and efficient decentralized storage solutions.