CryptoMaze Quantum Stress Test: Will Private Multi-Path Off-Chain Payments Survive Post-Quantum Cryptography?

04.11.2026

CryptoMaze is a classic atomic multipath payment protocol for payment channel networks. It solves an important and elegant engineering problem: how to split a large payment into parts without creating unnecessary off-chain contracts on shared edges and preventing intermediaries from linking the payment parts. However, its cryptographic foundation is the elliptic curve discrete logarithm and ECDSA-like scriptless locks. Therefore, the honest scientific question is not “Is CryptoMaze broken today?” but rather:  what exactly in the PCN architecture will have to be rebuilt if ECDLP and ECDSA become unreliable?

Keywords: Payment Channel Networks, Lightning Network, CryptoMaze, atomicity, unlinkability, wormhole attack, ECDLP, ECDSA, post-quantum cryptography, UC security, Bitcoin.

Content

  1. Intrigue: The Privacy and Quantum Transition Paradox
  2. What exactly does CryptoMaze do?
  3. Why is this an article about quantum technology?
  4. Research questions to be tested
  5. Evidence from the CryptoMaze article
  6. Linking multiple works into a coherent study
  7. Deep analysis: four theses
  8. Hypothesis and criterion for its refutation
  9. Limitations and fairness of findings
  10. What experiments are needed next?
  11. Conclusion
  12. Sources

1. Intrigue: The privacy paradox and quantum transition

⚙️ Detailed research option

Payment channel networks suffer from a subtle conflict. The larger the payment, the more difficult it is to find a single route with sufficient liquidity. Therefore, the payment is split into parts and routed along multiple paths. But as soon as multiple paths appear, three classic problems arise:  atomicity  (all parts must either pass through or fail to pass through together),  wormhole protection  (intermediaries must not steal fees by “skipping” honest participants), and  unlinkability  (colluding nodes must not realize they are seeing parts of a single payment).

CryptoMaze offers an elegant solution: first, a set of paths is mapped to a set of edges. Then, contracts are constructed not for each path individually, but rather edge-by-edge, in order from the sender. A single off-chain contract is created on shared edges, rather than several. This saves setup costs and reduces linking risk. However, this elegance conceals a cryptographic dependency: the contract terms are represented as group equations over an elliptic curve, while secret release and scriptless ECDSA lock rely on discrete logarithm complexity.

Hence the intrigue of the article:  CryptoMaze is only as good as its algebraic commitments.  If a future quantum adversary can solve ECDLP, the problem won’t just be with Bitcoin signatures. The very logic of conditional payment will be compromised: a commitment will cease to be a commitment, and “unlinkability” could become an illusion.

2. What exactly does CryptoMaze do?

⚙️ Detailed research option

According to the paper  “CryptoMaze: Privacy-Preserving Splitting of Off-Chain Payments”  (arXiv:2005.07574v3), the protocol solves the problem of atomic multipath payments in PCNs. A sender  U0 wants to transfer an amount  val to a recipient  Ur. Instead of independent HTLCs on each path, the protocol:

  1. constructs a set of channels  PC as the union of the edges of all selected paths;
  2. in the preprocessing phase, it creates conditions  R_{i,j} for off-chain contracts;
  3. uses  a 1-out-of-m  policy: if a node forks a payment, it only needs a release from at least one outgoing neighbor to avoid losing coins;
  4. ensures  non-linkability : different conditions on different outgoing channels prevent colluding neighbors from linking parts of the payment;
  5. uses timeouts that increase from receiver to sender and the UC model to describe security.

The authors claim the following properties: correctness, consistency, balance sheet security, value privacy, unlinkability, relationship anonymity, and atomicity. In experiments: approximately 11 seconds for a 0.04 BTC payment in a network of 25,600 nodes; worst-case communication costs less than 1 MB; in real Lightning Network snapshots, the average payment time is 0.39 seconds and 1.85 seconds for two snapshots; 50–60% savings on installing off-chain contracts on shared edges compared to the state-of-the-art.

For our topic, not only metrics are important, but also  cryptographic building blocks : ECDLP, hash function  H:{0,1}*→{0,1}^λ, blinding factor  e = H(x||id), group equations  R = e x G + R' + x' G, timelocks, and scriptless ECDSA lock. These become the points of the quantum stress test.

3. Why is this an article about quantum technologies?

⚙️ Detailed research option

The question constructor  quantum_research_questions_ru.html  suggests not mixing computation, networks, sensors, and cryptography, but rather separately testing the threat model, resources, standards, and migration costs for cryptographic scenarios (questions #49–56). This is a very useful framework. CryptoMaze isn’t a quantum protocol, but it serves as  an architectural stress test : what happens to a private multipath payment when a cryptographic assumption changes?

Historical background: in August 2024, NIST approved FIPS 203, 204, and 205—standards for post-quantum key establishment and digital signatures; the NCCoE is separately considering practical migration. This doesn’t mean the “quantum apocalypse” has already arrived. It does mean that designing long-lived protocols without analyzing the post-quantum replacement of primitives is already risky.

An important distinction.  Shor’s algorithm threatens factorization and the discrete logarithm—and therefore ECDLP and ECDSA. Grover’s algorithm provides a quadratic search speedup, so hash functions with  λ security bits need to be designed with a safety margin. But the main question with CryptoMaze isn’t the hash length, but whether it’s possible to replace  the algebraic conditional payment chain  with a post-quantum construction while preserving atomicity and unlinkability.

In the CryptoMaze paper, UC security is proven in a hybrid world with ideal blockchain and secure messaging functionality. This is a classical computing model. A post-quantum adversary requires, at a minimum, a revision of the assumptions and, possibly, a quantum UC formulation. You can’t simply say “replace ECDSA with ML-DSA” and keep all the proofs: the commitment structure, message size, verification cost, and the release logic itself will change.

4. Research questions to be tested

⚙️ Detailed research option

Below is not a list of rhetorical questions, but a core set of testable questions selected from the constructor. For a focused article, 12-18 general and 4-8 specific questions are sufficient; here, we focus on the cryptographic profile and practical migration.

No.Question to be testedWhat is considered evidence?Risk of false conclusion
Q1What specific cryptographic problem underlies CryptoMaze’s security?Explicit reduction to ECDLP/ECDSA and hash assumptions in contract descriptions and scriptless lock.Mix up “ECDSA is used” and “all security is in ECDSA only”.
Q2What exactly will a quantum adversary break: a signature, a commitment, a timelock, or the entire protocol?A step-by-step analysis of the preprocessing, forwarding, and release phases.Reduce the threat to just the Bitcoin signature.
Q3Will atomicity be preserved when replacing ECDSA with post-quantum signature?Formal verification: can an intermediary release a secret without satisfying a chain of conditions?Accept signature replacement as lock primitive replacement.
Q4Will 1-out-of-m balance sheet security be preserved?Checking equations  R_{i,j} and release from one of the neighbors.Lose branching when moving to another algebra.
Q5Will unlinkability remain with a quantum adversary?Can the adversary compute discrete logarithms and relate  R_{M,A},  R_{A,B},  R_{A,C}.Consider hash masking sufficient without group equations.
Q6Is it enough to increase the hash length?Grover threat analysis  H(x||id) and commitment.Ignore the Shor threat to ECDLP.
Q7What are the overhead costs of post-quantum migration?Size of signatures/proofs, number of messages, verification time.Compare only crypto primitives, not the entire protocol.
Q8Is it possible to preserve the “many paths → many edges” architecture?Check whether contract savings depend on EC algebra.Throw away a useful engineering idea along with a crypto-primitive.
Q9How will the UC proof change?Explicitly state hybrid functionalities and assumptions about the post-quantum adversary.Carry over the classical proof without changes.
Q10Which works from the CryptoMaze article form the sequence “HTLC → AMHL → AMP → Boomerang → Eckey → CryptoMaze”?Comparison table of atomicity, wormhole, linkability and shared edges.Mix different threat models and metrics.
Q11Is there a post-quantum analogue of scriptless ECDSA lock?An overview of PQ signatures, commitments, and zk approaches; checking compatibility with PCN.Declare “an analogue exists” without a construct.
Q12What is the minimum demonstration for a PCN user?Prototype with PQ-lock, measuring success, time and communication.Promising “quantum security” with just one signature.
Q13Which statements are an assessment of future risk, and which are an experiment today?Separation of facts, assessments and forecasts in the text.Create a sensation out of a hypothetical attack.
Q14What new question arises from the comparison of CryptoMaze and post-quantum standards?Formulate a testable hypothesis about the PQ-lock for PCN.Leave the article without the next step.
Q15What limitations make the conclusion preliminary?No PQ benchmark, classic UC model, experimental secp224r1.Hide restrictions for the sake of a nice title.

5. Evidence from the CryptoMaze article

⚙️ Detailed research option

Below are facts that can be used as evidence. Important: this evidence is for  classical  security and performance, not post-quantum security and performance.

Approval / mechanismWhat the article saysWhy is this important for quantum analysis?
PCN modelBipartite graph  G=(V,E), capacity function, off-chain contracts, HTLC as an example.Shows that payment security is distributed across channels and contracts.
Wormhole attackHTLC uses one commitment per path; colluding nodes can “jump” honest intermediaries and steal the commission.Atomicity and consistency are not bureaucracy, but protection against theft.
Many edges instead of pathsPaths  can p1 also  p2 share channels; instead of two contracts on a common edge, one is created.Savings of 50-60% on installation costs; the architectural idea can survive the change of primitives.
1-out-of-m policyIf a node forks a payment, it only needs a release from one outgoing neighbor.Balance sheet security depends on the algebra of obligations; under PQ-lock, this must be proven anew.
Contract termsR_{i,j}=e_{i,j}x_jG+R_{j,k} and the option with  +x_{j,k}G branching.These are group equations over an elliptic curve – a direct target of the Shor threat.
UnlinkabilityDifferent conditions apply to different channels; collusion between neighbors should not link payment parts.If DLP is lightweight, the connection itself  R can become computable.
Timeoutst_{i,j}=max{t_{j,k}}+Δ; timeouts grow towards the sender.Timelocks do not protect against forgery of an obligation if the release can be calculated.
UC modelPerfect functionality  F in a hybrid world  (F_B,F_smt).Classic proof; a new model is needed for the PQ adversary.
Scriptless ECDSA lockIdeal functionality  F_ECDSA-Lock; signature is used as lock.Direct dependence on ECDSA/ECDLP; signature replacement does not automatically preserve lock logic.
Experimentssecp224r1, OpenSSL, 25,600 nodes, 11s, <1MB, average 53.18KB and 93.203KB.The metrics are useful as a baseline for the future PQ prototype.

6. Linking multiple works into a coherent study

⚙️ Detailed research option

To avoid this article looking like a “parsing of a single PDF,” we’ll link the work that CryptoMaze relies on into a line of development: from simple HTLC to anonymous multi-hop locks, then to atomic multi-path payments, then to attempts to reduce latency, and finally to CryptoMaze with shared edge optimization.

Work / approachWhat decidesWhat remains a problemContact CryptoMaze
HTLC / LightningSingle-path off-chain payment, hash lock + time lock.Doesn’t scale well to large amounts; one commitment per path.Baseline; CryptoMaze generalizes to the multipath case.
Multi-Hop HTLC / AMHLPrivacy and anonymous multi-hop locks.Complex ZK proofs; does not solve multi-way atomicity automatically.Shows the price of privacy and anonymity.
AMPAtomic multi-path payment via secret shares.Wormhole attack; same commitment on paths.CryptoMaze compares to AMP in terms of time and communications.
BoomerangReduce latency and increase throughput.Wormhole and redundant collateral.Shows that atomicity is not enough – protection against collusion is needed.
Eckey et al.Local dynamic payment splitting, homomorphic encryption.High computational cost; recipient’s public key may facilitate linking.Alternative philosophy: don’t predetermine the path.
CryptoMazeAtomicity, wormhole-proof, unlinkable, single contract per shared edge.Classical EC algebra; no post-quantum analysis.Our case study for quantum stress testing.

The research flow is as follows:  HTLC  imposes a baseline conditional payment;  AMHL/Multi-Hop HTLCs  enhance privacy but are expensive;  AMP  makes multipath payments atomic but vulnerable to wormholes;  Boomerang  attacks latency but doesn’t solve all collusion problems;  Eckey et al.  try to avoid hard-coded paths;  CryptoMaze  optimizes shared edges and unlinkability. The next step isn’t “another protocol,” but a test:  which of these ideas survive post-quantum primitive replacement?

7. Deep analysis: four theses

⚙️ Detailed research option

Thesis 1. The quantum threat to CryptoMaze is not just a signature, but a commitment

⚙️ Detailed research option

In the classic description of CryptoMaze, a node receives a condition  R_{i,j} and must produce a discrete logarithm to claim the coins. This isn’t just a “transaction signature.” It’s  an algebraic commitment :  R = e x G + R' + x' G. If ECDLP becomes computable, an adversary can attempt to compute the required discrete logarithms and release the commitment outside the honest chain. Then, not only the signature cryptography suffers, but also  consistency  and  balance sheet security .

Thesis 2. Replacing ECDSA with a post-quantum signature is a necessary but not sufficient condition

⚙️ Detailed research option

You can replace ECDSA with ML-DSA or SLH-DSA for message authentication. However, scriptless locking in CryptoMaze uses a signature as  a conditional release mechanism . If you simply add a PQ signature next to it, the equations  R_{i,j} will still remain EC equations. We need either a post-quantum equivalent of scriptless locking or replacing the entire conditional construction with a PQ commitment + zk-proof/aggregable signature. This dramatically changes message size and verification costs.

Thesis 3. CryptoMaze’s unlinkability also relies on the complexity of DLP

⚙️ Detailed research option

The authors emphasize that different conditions on different channels prevent colluding neighbors from linking parts of a payment. However, if an adversary can calculate discrete logarithms, they can analyze the relationships between  R_{M,A},  R_{A,B} and  R_{A,C} in ways different from those assumed in the classical model. Hash obfuscation  e=H(x||id) helps, but it doesn’t eliminate group algebra. Therefore, post-quantum unlinkability is a separate research problem, not a free bonus for a new signature.

Thesis 4. The engineering idea of ​​”paths → edges” can survive the change in cryptography

⚙️ Detailed research option

The most practical part of CryptoMaze is the mapping of multiple paths to multiple edges and a single contract on a shared edge. This idea reduces setup costs and reduces the number of objects that need to be linked. It doesn’t have to die with ECDLP. Even if the conditions become PQ proofs, we can preserve:

  • merging edges instead of duplicating contracts;
  • timeouts growing towards the sender;
  • 1-out-of-m policy for branching;
  • separation of the preprocessing/forwarding/release phases.

But each of these properties will have to be proven anew in the post-quantum model. This is where the interesting scientific contribution lies: not “CryptoMaze is not quantum-secure”—that’s too simplistic—but  which architectural invariants of CryptoMaze are independent of ECDLP, and which are a consequence of it .

8. Hypothesis and criterion for its refutation

⚙️ Detailed research option

Hypothesis H.  Post-quantum migration of private multi-path off-chain payments does not boil down to replacing ECDSA with a standardized PQ signature. The CryptoMaze architecture, in which atomicity and non-linkability are tied to group equations over an elliptic curve, requires a new post-quantum lock primitive; without it, the properties of balance sheet security, consistency, and non-linkability degrade even with formally “post-quantum” signatures.

Verification criterion:  Build a PQ-CryptoMaze prototype with an explicitly defined post-quantum lock primitive and prove in a suitable UC model that:

  1. atomicity is preserved in the case of collusion of intermediaries;
  2. 1-out-of-m balance sheet security is maintained;
  3. non-linkability is no worse than the classical model;
  4. Communication and computational costs remain within practically acceptable limits.

What will disprove the hypothesis?  If it can be shown that simply replacing the signature and hash with PQ variants without changing the algebra of the conditions preserves all the listed properties and the UC proof carries over without significant changes, the hypothesis will be disproved or significantly refined. Currently, the CryptoMaze paper does not contain such proof.

9. Limitations and fairness of findings

⚙️ Detailed research option

  • There is no quantum computer experiment.  We are discussing a threat model and an architectural vulnerability, not a demonstration of an attack.
  • There is no PQ benchmark.  CryptoMaze metrics (11 s, <1 MB, 0.39 s, 1.85 s) are specific to the classic implementation and are not automatically transferable to the PQ version.
  • The UC proof is classical.  It does not cover quantum adversaries and quantum channels.
  • secp224r1 in experiments.  This is convenient for evaluating performance, but it is not a production curve for the Lightning Network; security and performance in the real network may vary.
  • Not all works in the article are equal in their threat modeling.  AMP, Boomerang, Eckey et al., and CryptoMaze solve different, related problems; they should be compared using comparable metrics.
  • Post-quantum standards are not a complete protocol.  FIPS 203/204/205 provide the primitives, but do not solve the problem of conditional payment in PCNs.

10. What experiments are needed next?

⚙️ Detailed research option

  1. Formalization of the PQ model.  Describe an adversary who solves ECDLP and has access to quantum resources; reformulate the ideal payment functionality.
  2. Candidate PQ-lock.  Explore combinations of PQ signatures, commitments, and zk-proofs that preserve a one-out-of-m release.
  3. Prototype.  Implement at least one path and one branch with PQ primitives; measure message size, verification time, and the number of rounds.
  4. Baseline comparison.  Compare with classic CryptoMaze in terms of TTP, communication overhead, and contract setup costs.
  5. Verifying non-linkability.  Simulate a collusion between neighbors and check whether the linking of payment parts creates new obligations.
  6. Negative result.  If PQ-lock is too expensive, this is also an important result: it will show the limits of the architecture’s applicability.

A minimal demonstration that will interest both a researcher and an engineer:  the same multipath payment with a shared edge, implemented in both the classic and PQ variants, with a fair comparison of time, traffic, and preservation of atomicity under collusion.

11. Conclusion

⚙️ Detailed research option

CryptoMaze is a strong classical protocol. It demonstrates how engineering care—combining paths into edges, one-out-of-m release, careful timeouts—can simultaneously improve atomicity, wormhole resistance, and unlinkability. But its cryptographic foundation is not post-quantum. The main scientific question is not whether CryptoMaze is broken, but  which of its properties are architectural and which are a consequence of ECDLP .

If future work constructs a post-quantum lock for PCNs and proves that atomicity, balance security, and unlinkability are preserved with acceptable overhead, this will be a strong result. If not, we will reach an equally compelling conclusion: private multi-path off-chain payments will require a more thorough redesign than simply replacing the signature. This testable question is the next step in this research.

12. Sources

⚙️ Detailed research option

  1. Mazumdar S., Ruj S.  CryptoMaze: Privacy-Preserving Splitting of Off-Chain Payments . arXiv:2005.07574v3. In the article: PCN, atomic multipath payment, wormhole attack, unlinkability, UC model, scriptless ECDSA lock, experimental metrics.
  2. quantum_research_questions_ru.html . “Quantum Technologies: 75 Questions for a Research Article.” Methodological framework: problem statement, mechanism, evidence, cryptography, migration, limitations.
  3. National Quantum Initiative. Quantum science areas: computing, networks, sensors.  quantum.gov .
  4. NIST. FIPS 203, 204, 205 – Post-quantum key establishment and digital signature standards, approved August 13, 2024.  csrc.nist.gov .
  5. NIST NCCoE. Migration to Post-Quantum Cryptography.  nccoe.nist.gov .
  6. Poon J., Dryja T. The Bitcoin Lightning Network: Scalable Off-Chain Instant Payments, 2016.
  7. Malavolta G. et al. Anonymous Multi-Hop Locks for Blockchain Scalability and Interoperability, NDSS 2019.
  8. Osuntokun O. AMP: Atomic Multi-Path Payments over Lightning, 2018.
  9. Bagaria V., Neu J., Tse D. Boomerang: Redundancy Improves Latency and Throughput in Payment-Channel Networks, 2020.
  10. Eckey L. et al. Splitting Payments Locally While Routing Interdimensionally, IACR ePrint 2020/555.

Pre-publication verification.  Observations, estimates, and predictions are separated; the existence of a ready-made quantum attack on CryptoMaze is not claimed; classical metrics are not presented as post-quantum; model limitations are indicated; a new test question and a subsequent experiment are formulated.