ESSPI: ECDSA/Schnorr Signed Program Input for BitVMX
Sergio Demian Lerner, Martin Jonas, Ariel Futoransky · 2025
Read ESSPI: ECDSA/Schnorr Signed Program Input for BitVMX by Sergio Demian Lerner, Martin Jonas, Ariel Futoransky (2025) 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-2503-02772.txt.
Introduction
Sergio Demian Lerner, Martin Jonas, and Ariel Futoransky introduce ESSPI, a way to sign program inputs for BitVMX with ECDSA or Schnorr signatures. Their 2025 paper deals with Bitcoin-based protocols that have long used one-time signatures.
Schemes such as Lamport and Winternitz carry large storage overheads, which limit practical use. With Schnorr signatures, ESSPI reaches a one-to-one data expansion, far better than the best Winternitz ratio. Four innovations make this work. They add a challengeable hashing core to the BitVMX CPU. They also add a partition-based search to detect fraud during hashing. An enhanced transaction graph and a timelock method for proving data availability complete the set.
Catalog record
| CREATOR | Sergio Demian Lerner, Martin Jonas, Ariel Futoransky |
|---|---|
| YEAR | 2025 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 79 KB |
| VERSION | 1.0 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2503-02772
Browse the full library · About the Tower of Babel
Full text
The opening of the work, from the library's hosted text:
ESSPI: ECDSA/Schnorr Signed Program Input for BitVMX Sergio Demian Lerner, Martin Jonas, Ariel Futoransky Abstract The BitVM and BitVMX protocols have long relied on inefficient one-time signature (OTS) schemes like Lamport and Winternitz to sign program inputs. These schemes exhibit significant storage overheads that hinder their practical application. This paper introduces ESSPI, an optimized method that utilizes ECDSA / Schnorr signatures to sign the input of the BitVMX program. With Schnorr signatures we achieve an optimal 1:1 data expansion, compared to the current known best ratio of 1:200 based on Winternitz signatures. To accomplish this, we introduce 4 innovations to BitVMX: (1) a modification of the BitVMX CPU, adding a challengeable hashing core to it, (2) a new partition-based search to detect fraud during hashing, (3) a new enhanced transaction DAG with added data-carrying transactions with a fraud-verifying smart-contract, and (4) a novel timelock-based method for proving data availability to Bitcoin smart contracts. The enhanced BitVMX protocol enables the verification of uncompressed inputs such as SPV proofs, NiPoPoWs, or longer computation integrity proofs, such as STARKs. 1 Introduction Several schemes exist today to optimistically verify Bitcoin computations, such as BitVM2 [ 1 , 2 ] , BitVMX [ 3 ] , SNARKnado [ 4 ] , and BitSNARK [ 5 ] . From these, BitVMX stands out for being the most efficient in terms of onchain cost of dispute, the simplicity of its design, and the generality of the programs it can run without relying on SNARKs. One of the key limitations of BitVMX is that program inputs need to be signed with inefficient OTS schemes such as Lamport or Winternitz. The Winternitz scheme, which is the most efficient of the two, requires on average 25 witness bytes (or vbytes [ 6 ] ) per signed bit. Unlike standard schemes like ECDSA or Schnorr, which have constant size signatures, Lamport/Winternitz signatures and their public keys grow proportionally with the size of the data to be signed. One of the key open questions related to BitVM protocols is whether there is a more efficient method to sign a BitVM program input. Although the advantage is less significant for BitVM2 that only accepts 300-byte SNARK proofs as input and consumes megabytes in midstates, it is highly relevant to generic verification protocols such as BitVMX that can check not only SNARKs but also larger proofs produced by Nova ( \sim 10 Kbytes) or STARK ( \sim 80 Kbytes). Last, with public-key signed program inputs, BitVMX programs can verify uncompressed inputs, such as BLS signatures for payment channels or NiPoPoWs inputs for bridges. The next question is: Can we use a standard signature scheme to sign the program input, benefiting from a constant short signature? Using a standard number-theoretic public-key scheme would not only reduce the on-chain footprint, but also avoid storing and transferring the many one-time public keys required by current BitVM protocols. The benefit is even greater when the BitVMX protocol is extended to n parties, because the number of one-time public keys needed increases quadratically with n. In this paper, we positively answer all these previously open questions with a new variant of the BitVMX: ESSPI. The main trick in ESSPI is to use Bitcoin transactions to carry the program input as payload. We sign the hash of the program input twice, once with a public-key scheme and another time with Winternitz, and we prove the equivalence of these two signatures in a second BitVMX instance. This solves the problem of verifying the program input authenticity in the BitVMX CPU. However, because the Winternitz signature only signs the hash of the input, and not every input bit, it creates a new problem: ensuring input data availability. Although embedding signed data in the blockchain can be used to prove data availability and data authenticity to the full nodes, we still need to build a proof mechanism to convince a BitVMX smart contract of data availability. We present another trick to prove the availability of a transaction to a predefined smart contract where the transaction ID is unknown to the smart contract at the time of creation. Figure 1: The method (a) to authenticate the program input used in the current version of the BitVMX protocol requires long Winternitz public keys and publishes an encoded version of the full program input along the signature resulting in long transaction witnesses. The improved method (b) authenticates the input using a hash digest but does not prove the program input data availability because it doesn’t publish the program input inside the witness, only its short digest. The paper is organized as follows. The first section discusses the main problem: how to provide public-key-authenticated data to be used as program input for the BitVMX protocol. The next section lists different ways to use Bitcoin as an efficient data availability layer for large data chunks using standard transactions. In the last section, we show how to build DA proofs specifically for the BitVMX protocol. 2 Overview of BitVMX BitVMX is an optimistic proving system that allows the verification of arbitrary computations on Bitcoin. Using BitVMX, two parties can create a dispute resolution game on-chain, where they take the respective roles of prover and verifier. The prover can then submit the proof for some arbitrary computation, while the verifier can challenge the proof in case of disagreement. This proof consists of the input and final state of a virtual CPU after executing a pre-agreed program. To ensure that both parties commit to using the BitVMX mechanism, they must both lock some bitcoin as part of the setup process, creating a UTXO whose spending is restricted by the outcome of the protocol. The dispute resolution game consists of multiple on-chain rounds where the verifier challenges specific parts of the data that the prover has provided, and the prover must respond with the requested information. The game is designed such that if the proof is invalid, the prover will be forced to commit to conflicting data across different rounds. This results in the prover being unable to respond to at least one challenge, as no valid transaction could be created satisfying the constraints established by the previous transaction scripts. Each round has an associated time lock, and if either party fails to respond within this timeframe, they lose the game, and the other party can claim the funds previously locked. This type of system is called optimistic because the prover wins after a predefined time period if the verifier does not challenge the initial proof, leading to almost no interaction with on-chain transactions. The dispute resolution game is implemented as a directed acyclic graph (DAG) of interconnected transactions, where BitVMX emulates covenants using transactions presigned by all the parties involved. A message-linking scheme based on Winternitz signatures ensures that both parties sign their messages, and relative timelocks are used to force the parties to engage in the game once it is kicked off. Winternitz signatures are also used to authenticate the input of the program. One of the distinguishing properties of BitVMX is that it verifies the memory consistency using an execution step hash chain instead of using memory Merkelization. When the protocol finds a disagreement step in the execution trace