Catching the Fly: Practical Challenges in Making Blockchain FlyClient Real

Pericle Perazzo, Dario Capecchi · 2026

Read Catching the Fly: Practical Challenges in Making Blockchain FlyClient Real by Pericle Perazzo, Dario Capecchi (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2604-26736.txt.

Introduction

Perazzo and Capecchi bridge theory and deployment for FlyClient, a lightweight blockchain verification protocol enabling proof-of-work validation with minimal data for wallets, Internet-of-Things devices, and cross-chain bridges. Despite its promise, FlyClient remains experimental, with few real deployments and performance tests. The paper offers a new adversary model that parametrizes a verifier under a concrete economic reading while saving proof space. It also gives the first practical FlyClient prover for a production blockchain in Zcash. Two optimizations shrink proof size, the first needing no consensus change.

Catalog record

CREATORPericle Perazzo, Dario Capecchi
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE61327
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2604.26736

Browse the full library · About the Tower of Babel

Full text

The opening of the work, from the library's hosted text:

Catching the Fly: Practical Challenges in Making Blockchain FlyClient Real Pericle Perazzo, Dario Capecchi Abstract FlyClient is a lightweight blockchain verification protocol that enables proof-of-work validation using minimal data, making it ideal for resource-constrained environments like mobile wallets, Internet-of-Things devices or cross-chain bridges implemented with smart contracts. Despite its strong potential for enabling lightweight blockchain verification, FlyClient protocol is still in the experimental stages, with limited real-world deployments and performance evaluations under diverse conditions. In this paper we bridge the gap between theory and deployment, by addressing several technical challenges to advance FlyClient to a production-ready solution. Namely, our contribution is three-fold: (i) we formally introduce an adversary model alternative to the original FlyClient one, that allows us to parametrize a verifier under a concrete economic interpretation, while also saving some proof space; (ii) we provide the first practical FlyClient prover implementation for a production blockchain (Zcash), and we estimate its performance under different configurations; (iii) we introduce and evaluate two optimizations that minimize the size of FlyClient proofs, the first of which does not require any consensus change. Submitted to IEEE Transactions on Dependable and Secure Computing I Introduction FlyClient [ 4 , 12 , 13 ] is an innovative protocol designed to enable ultra-light clients for Proof-of-Work (PoW) blockchains. Traditional blockchain clients require downloading and verifying large amounts of data to ensure security and correctness, which is impractical for resource-constrained execution environments. FlyClient addresses this challenge by leveraging probabilistic sampling of block headers combined with Merkle Mountain Range (MMR) commitments, thus producing proofs of chain validity that grow only logarithmically with the chain length. The FlyClient protocol allows clients to verify the entire blockchain with minimal bandwidth and storage requirements, while maintaining strong security guarantees. It can produce both interactive or non-interactive proofs, and it can also deal with PoW difficulty variability. FlyClient finds application in those cases in which the state of the blockchain must be verified by execution environments with limited resources. A straightforward application is mobile wallets, which must make sure the presence of given transactions inside the blockchain. Current mobile wallets trust third-party full nodes to maintain and check the whole blockchain, but FlyClient can do the same without any trust assumption. Another application is in the Internet of Things (IoT), in which resource-constrained devices can use FlyClient to perform machine-to-machine transactions or read and write immutable logs. A third application is in blockchain bridges, which are aimed at moving a crypto-asset from a blockchain to another blockchain and back. Current blockchain bridges employ trusted third-party nodes or federations of nodes, but FlyClient allows a smart contract to verify the state of a foreign blockchain by validating a non-interactive FlyClient proof. This permits trustless bridges, in which the correctness of a cross-chain interaction is automatically proved without trusting any (particular) node. A blockchain can support the FlyClient functionalities natively [ 12 ] or by means of smart contracts [ 13 ] . Among the cryptocurrencies that natively support FlyClient, Zcash [ 8 ] is the most important and capitalized one. Zcash integrated FlyClient through ZIP 221 [ 3 ] as part of the Heartwood network upgrade. Despite this, no FlyClient prover for Zcash has been developed yet, and the FlyClient protocol is still in the experimental stages. This paper addresses several technical challenges to advance FlyClient from a theoretical concept to a production-ready solution. In particular, its contribution is three-fold. First, we introduce the w_{a} -adversary model as an alternative to the original FlyClient’s (c,L) -adversary model. The w_{a} -adversary is based on a single parameter that directly captures the budget available to the attacker for performing his attack, and thus it has a concrete economic interpretation useful for real deployments. A FlyClient verifier can be parametrized by assuming a certain w_{a} -adversary and solving an optimization problem, which also offers some proof size reduction that we measure. Second, we present our experiences in extending Zebrad [ 17 ] , which is the Rust-based official full node implementation of Zcash, for offering FlyClient prover functionalities. We also estimate the extra resources that are needed for such functionalities, and we measure the proof size under different representations and different verifier variants. The prover implementation is made available to the Zebrad community, which, at the time of writing, has already accepted some of our extensions in the Zebrad official repository. Third, we introduce and evaluate two optimizations that aim at minimizing the size of the interactive and non-interactive FlyClient proofs. Namely, we introduce the cumulative proofs and the distilled proofs . The former optimization offers limited advantage (-9.16% non-interactive proof size), but it is already deployable at zero cost. The latter one requires a change in the consensus layer, but it offers a more substantial advantage (-71.33% non-interactive proof size). The rest of the paper is structured as follows. Section II discusses related work. Section III explains the necessary background concepts. Section IV introduces the w_{a} -adversary model and the relative configuration method for the verifier. Section V introduces and evaluates our FlyClient prover implementation for Zcash. Section VI introduces and evaluates our proposed FlyClient proof optimizations. Finally, the paper is concluded in Section VII . II Related Work II-A Research on FlyClient Bünz et al. [ 4 ] introduced the FlyClient protocol to enable extremely lightweight verification of PoW blockchains by reducing the data requirements for clients from linear to logarithmic complexity. FlyClient achieves efficiency through probabilistic sampling and MMR commitments embedded in block headers, allowing clients to verify the entire chain with only a logarithmic number of sampled blocks, even in the case of variable-difficulty blockchains. The authors model the adversary as a (c,L) -adversary, that is an attacker that cannot produce an L -long fork with c validity ratio. Although this modeling is convenient to mathematically prove the security properties of FlyClient, it lacks a direct economic interpretation, thus it is less convenient from the point of view of a real deployment. The present paper introduces the alternative w_{a} -adversary model, which is based on a single parameter that directly captures the budget available to the attacker for performing his attack, and thus it has a concrete economic interpretation. Nemoz and Zamyatin [ 12 ] explore the security implications of deploying FlyClient using a velvet fork, which allows backward compatibility without requiring all nodes to upgrade. In particular, the authors explored the possibility of chain-sewing attacks (originally introduced by [ 11 ] ), in which an adversary stitches together parts of different chains to deceive FlyClient verifiers about the state of the blockchain. This attack is feasible because, in a velvet fork, a block could contain invalid FlyClient data without being rejected by the nodes. However, this does not apply to Zcash, which implemented FlyClient via a hard fork. In the Zcash case, all blocks are required to include valid FlyClient data, hence preventing the conditions necessary for the chain-sewing attack to succeed. Our FlyClient prover implementation is therefore immune to chain-sewing attacks. Perazzo and Xefraj [ 13 ] proposed SmartFly, a FlyClient implementation via smart contracts which does not require forks at all. The nodes that invoke the smart contract methods do not need to be trusted, as long as the smart

Read the complete text