BioZero: Privacy-Preserving and Publicly Verifiable On-Chain Biometric Authentication via Homomorphic Commitments and Zero-Knowledge Proofs

arXiv:2409.17509 · cs.CR · Submitted 2024-09-26 · Read on arXiv

Listen

Radio episode about this paper

Transcript

Introduction to the show: ident: Security Radio. Generated commentary on the latest security and cryptography papers.

Nadia: I'm Nadia.

Elias: And I'm Elias, here with Priya.

Nadia: We're looking at the paper: "BioZero: Privacy-Preserving and Publicly Verifiable On-Chain Biometric Authentication via Homomorphic Commitments and Zero-Knowledge Proofs".

Elias: Here's the gist of the paper. Decentralized identity systems promise user-controlled identifiers and cross-domain verification without a shared identity provider, yet authentication still collapses to possession of keys or credentials once secrets are leaked, reused, or replayed <ref:2409.17509#pg2>.

Nadia: BioZero presents a privacy-preserving biometric authentication protocol for decentralized identity settings that enables publicly verifiable on-chain authentication decisions while keeping biometric templates hidden <ref:2409.17509#pg2>.

Elias: Protocol Core and Cryptographic Tools BioZero combines commitment-homomorphic computation over Pedersen commitments with Groth16 zero-knowledge proofs to realize identity-bound authentication with succinct on-chain verification <ref:2409.17509#pg3>.

Nadia: The protocol utilizes Pedersen commitments to protect biometric templates while enabling distance computation over commitments, and it uses Groth16 zk-SNARKs to provide succinct, noninteractive proofs with efficient verification suitable for onchain checks <ref:2409.17509#pg3>.

Paper summary: Nadia: So, to wrap up on "BioZero: Privacy-Preserving and Publicly Verifiable On-Chain Biometric Authentication via Homomorphic Commitments and Zero-Knowledge Proofs," the paper presents a protocol that successfully couples Pedersen commitment based homomorphic checks with a compact Groth16 proof.

Elias: That combination allows for public verifiability without revealing the biometric templates, which they claim is their main achievement in creating this identity bound authentication protocol.

Priya: And from my side, it seems like the real value here is that they’ve shown a way to have a decentralized system where you can verify an identity using biometrics while maintaining template privacy through these specific cryptographic mechanisms.

Nadia: It’s about making those on-chain authentication decisions possible without relying on a single shared identity provider, which is what the initial problem was driving this research.

Elias: The paper moves beyond just claiming it can work by providing a thorough security analysis around acceptance soundness, template privacy, freshness, and non-malleability under an open threat model.

Priya: And it’s also very interesting how they managed to balance the need for strong privacy with practical performance gains on a blockchain, even dealing with the quantization trade-offs in their implementation.

Nadia: So it’s not about a single magic solution but a practical authentication layer that fits into existing decentralized identity settings, and the results support that idea.

Elias: It shows how commitment-based homomorphic checks combined with zero knowledge proofs can be used to create something publicly verifiable without exposing sensitive biometric information.

Conclusion: Nadia: So, we're wrapping up on BioZero today, which is about using these commitment and zero knowledge proof tricks to do biometric authentication on a decentralized identity system without actually showing anyone your template.

Elias: Yeah, the title says it's privacy-preserving and publicly verifiable. It’s trying to solve that core problem where you can't have a shared identity provider but still need secure authentication.

Priya: What I see is they’re using Pedersen commitments to hide the biometric data itself while letting you do computations on those hidden values, and then they layer a Groth16 zero knowledge proof on top of it.

Nadia: Exactly, and what's cool is that this whole thing allows for verification right on the blockchain, which means anyone can check if someone is who they say they are without ever seeing the actual biometric sample.

Elias: The authors focus heavily on five properties like acceptance soundness and template privacy, showing how their math holds up under an open threat model. They link template privacy directly to the zero knowledge property of that Groth16 proof.

Priya: But I'm still looking at the practical stuff, like how they handle real-world data. They have to do some quantization on these integer messages for the commitments, and they show accuracy loss is really small, under one percent even with moderate settings like a factor of sixty-four or one hundred twenty-eight.

Nadia: That’s key because it means this isn't just a theoretical gadget; they found a practical operating point that keeps the privacy strong while keeping the biometric matching accurate enough for real use.

Elias: And on the performance side, they report some big speedups, up to six times faster in total authentication latency compared to a standard setup, which is pretty significant for on-chain stuff.

Priya: It’s important to remember though that BioZero authenticates the witness vector you send from your device pipeline; it doesn't verify if the sensor capture was actually fresh or if the person is alive right now.

Nadia: Right, so it’s a strong authentication layer for a decentralized system, but it doesn't cover the whole identity lifecycle—things like how you issue or revoke that biometric credential—that’s still outside this protocol’s scope.

Elias: It positions itself as an authentication tool rather than a full identity solution. This means its security relies on the integrity of the blockchain itself and standard math assumptions, not on solving all the messy problems of identity management.

Nadia: So, what does this mean for us right now? It shows how we can build useful decentralized systems where sensitive data stays private but still gets verified publicly through smart contracts.

Priya: It opens up possibilities for decentralized applications that need to prove a user has a certain trait without revealing the trait itself. That’s the direction this kind of protocol is heading.

Zhibin Lina, Taotao Wang, Junhao Laia, Shengli Zhanga, Qing Yanga, Soung Chang Liew

College of Electronics and Information Engineering, Shenzhen University · Department of Information Engineering, The Chinese University of Hong Kong

cs.CR

Submitted: 2024-09-26

Updated: 2026-11-01

Comments: 21 pages, 3 figures, 11 tables

License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/

Importance score: 83/100

The gist: Decentralized identity systems promise user-controlled identifiers and cross-domain verification without a shared identity provider…

Key concepts

Pedersen Commitments
These are cryptographic tools used to lock away biometric data (like templates) in a way that allows mathematical operations on the locked data without needing to see the actual data. They enable distance computations over these commitments while maintaining privacy.
Groth16 Zero-Knowledge Proofs
This is a type of cryptographic proof that allows someone to prove a statement is true (e.g., 'I possess a valid biometric match') without revealing the underlying secret information. Groth16 proofs are very compact and allow for quick verification on blockchains.
Homomorphic Computation
In this context, homomorphic computation means being able to perform calculations or comparisons on encrypted data (the commitments) directly. This allows the system to check if a user's input is close enough to their stored template without decrypting the template first.
Acceptance Soundness
This is a security property ensuring that if a user successfully passes the authentication check, they must genuinely possess the correct biometric witness. It relies on checks like commitment binding and zk-SNARK soundness to guarantee the result is trustworthy.

Terminology

Summary

Decentralized identity systems promise user-controlled identifiers and cross-domain verification without a shared identity provider, yet authentication still collapses to possession of keys or credentials once secrets are leaked, reused, or replayed <ref:2409. p. 017509#pg2>. BioZero presents a privacy-preserving biometric authentication protocol for decentralized identity settings that enables publicly verifiable on-chain authentication decisions while keeping biometric templates hidden <ref:2409. p. 017509#pg2>.

Protocol Core and Cryptographic Tools

BioZero combines commitment-homomorphic computation over Pedersen commitments with Groth16 zero-knowledge proofs to realize identity-bound authentication with succinct on-chain verification <ref:2409.17509#pg3>. The protocol utilizes Pedersen commitments to protect biometric templates while enabling distance computation over commitments, and it uses Groth16 zk-SNARKs to provide succinct, noninteractive proofs with efficient verification suitable for onchain checks <ref:2409. p. 117509#pg3>. The core authentication mechanism authenticates when the squared Euclidean distance is below a threshold epsilon <ref:2409. p. 217509#pg5>.

Security Properties and Analysis

BioZero aims to provide five protocol-level properties, including acceptance soundness, template privacy, freshness, and non-malleability under an open and decentralized threat model <ref:2409.17509#pg4>. Acceptance soundness reduces to commitment binding, transcript-bound consistency checks, zk-SNARK soundness, and the spot-check miss probability <ref:2409. p. 517509#pg6>. Template privacy is achieved through Pedersen hiding together with the zero-knowledge property of Groth16 <ref:2409. p. 517509#pg6>. Freshness and non-malleability are reduced to nonce monotonicity and Fiat–Shamir transcript binding <ref:2409. p. 517509#pg9>.

Performance Metrics and Efficiency Gains

The paper reports significant performance improvements on an Ethereum testbed, achieving up to 67.8× speedup in network-adjusted total authentication latency and up to 266. p. 04× speedup in client-side proving over a zk-SNARK-only baseline <ref:2409. p. 017509#pg2>. Verification runtime remains in the millisecond range (BioZero 28. p. 08 ∼ 41.2 ms vs. Vanilla ZKBio 35. p. 04 ∼ 77.6 ms) <ref:2409. p. 017509#pg2>. Total verification gas grows logarithmically with vector length, whereas the zk-only baseline grows linearly <ref:2409. p. 017509#pg2>.

Implementation Robustness and Quantization

The protocol operates over integer messages for Pedersen commitments, necessitating biometric embeddings to be quantized <ref:2409.17509#pg7>. Experiments on LFW models show accuracy loss below 1% across practical quantization ranges when using moderate quantization factors like k=64 or k=128 <ref:2409. p. 017509#pg11>. This demonstrates that BioZero provides a good operating point, balancing privacy-preserving integer processing with minimal accuracy degradation <ref:2409. p. 917509#pg11>.

System-Level Security Boundaries

BioZero authenticates the witness vector presented by the client-side acquisition pipeline, not biometric liveness or fresh sensor capture <ref:2409.17509#pg9>. The protocol does not address the full identity lifecycle, as issuance, revocation, recovery, template update, and uniqueness enforcement remain outside the protocol <ref:2409. p. 017509#pg9>. These boundaries are part of the intended scope of the construction <ref:2409. p. 017509#pg9>. The security arguments rely on the integrity of the underlying blockchain, correct contract execution, and standard cryptographic assumptions <ref:2409. p. 117509#pg9>.

Conclusion and Practical Application

BioZero is positioned as a practical authentication layer for decentralized biometric identity settings rather than a complete identity lifecycle solution <ref:2409.17509#pg13>. Its acceptance guarantee is defined with respect to the biometric witness encoded in the transcript rather than biometric liveness <ref:2409. p. 117509#pg13>. The protocol successfully couples Pedersen commitment-based homomorphic checks with a compact Groth16 proof to achieve public verifiability without revealing templates <ref:2409. p. 117509#pg8>. The results support BioZero as a practical authentication layer for decentralized biometric identity systems <ref:2409. p. 017509#pg2>.

Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.

Improvements for AI systems

  1. Decentralized identity binding for biometric authentication allows systems to bind an enrolled identifier id to a biometric witness without exposing template data, mitigating account takeover risks from stolen keys or credentials.

  2. The system enables publicly verifiable on-chain authentication decisions, meaning external applications can validate the authentication decision and its binding context directly from the on-chain evidence without trusting a centralized biometric verifier.

  3. The protocol combines commitment-homomorphic computation over Pedersen commitments with Groth16 zero-knowledge proofs to produce publicly verifiable authentication decisions without revealing biometric templates.

  4. Biometric authentication can be performed efficiently, achieving up to 266.4 times speedup in client-side proving over a zk-SNARK baseline, decoupling the dominant client-side cryptographic cost from vector length.

  5. The system supports quantization for integer messages in Pedersen commitments, allowing face recognition models to maintain high decision quality with accuracy loss below 1% across practical quantization ranges (e.g., k=64 for 128D embeddings).

Related papers