# QDBFT: A Dynamic Consensus Algorithm for Quantum-Secured Blockchain

Crypto.world

> Consortium blockchains built on PBFT-family protocols are fragile in two ways: their validator sets are usually static, and their security rests on public-key cryptography that Shor's algorithm can…

**arXiv**: 2602.11606

**Published**: 2026-02-12

**Authors**: Fei Xu, Cheng Ye, Jie OuYang, Ziqiang Wu, Haoze Chen, An Hua, Meifeng Gao, Qiandong Zhang, Minghan Li, Feilong Li, Yajun Miao, Wei Qi

## AI summary

## Overview

**Research area:** Blockchain consensus protocols and quantum-safe cryptography, specifically the integration of Quantum Key Distribution (QKD) networks with Byzantine Fault Tolerance (BFT) for consortium blockchains.

**Technical level:** Advanced. The paper assumes familiarity with PBFT-style three-phase consensus, consistent hashing, message authentication codes (HMAC/Toeplitz), and QKD key-pool concepts.

**Scope in one sentence:** The paper proposes QDBFT, a dynamic BFT consensus protocol that uses a consistent-hash-ring "Carousel" mechanism to rotate primary nodes through membership changes and replaces inter-node digital signatures with QKD-derived symmetric-key authentication tags.

## What This Paper Is About

Consortium blockchains built on PBFT-family protocols are fragile in two ways: their validator sets are usually static, and their security rests on public-key cryptography that Shor's algorithm can break. QDBFT attacks both problems at once — it lets nodes join, leave, or drop offline while consensus continues, and it secures the messages exchanged during consensus with keys produced by a QKD network rather than with classical signatures. The stated goal is to keep performance in the same range as traditional PBFT while removing the classical-cryptography attack surface from the core consensus path.

## Key Contributions

1. **Carousel: automatic primary-node rotation via a consistent hash ring.** Each node is assigned multiple virtual points on a 32-bit ring (range [0, 2³²−1]), and the current primary is chosen by mapping the parent block hash H(pb) onto the ring and taking the node owning the first hash point encountered clockwise. After a block commits, proposal rights shift to the next node determined by the same calculation, distributing validation authority equitably rather than concentrating it.
2. **Consensus-driven dynamic membership.** A configuration table T_v (with version v) is maintained collectively and updated to T^u_v only after consensus validation. Three event types are handled inside the protocol: node join (m_jo, confirmed after 2f consistent messages), voluntary exit (m_exit, processed after 2f−1 consistent messages), and unresponsive events (m_unresp, triggered after a survival probe m_sur gets no reply within threshold ΔT, then confirmed after 2f−1 consistent messages).
3. **QKD-based message authentication replacing signatures among consensus nodes.** Algorithm 1 has a sending node l compute a digest D(m) and, for every peer e, an authentication tag H′(k_{l,e}, D(m)) using directional pre-shared QKD keys, broadcasting the bundle m_α = {D(m), {H′(k_{l,e}, D(m))}}. Receivers verify with their local key and either advance the protocol or discard the message. H′ can be implemented with HMAC or the Toeplitz algorithm.
4. **A five-phase hybrid-cryptography consensus scheme.** REQUEST, NEW, TRANSMIT, COMMIT and REPLY phases, where client-to-node messages in REQUEST and REPLY use Post-Quantum Cryptography (PQC) signatures for user-facing compatibility, while inter-node messages in NEW, TRANSMIT and COMMIT use the QKD-based authentication to obtain information-theoretic security in the consensus core.

## Main Findings

- **Qualitative performance parity with PBFT:** The abstract states that experimental evaluations show QDBFT achieving performance "comparable to traditional PBFT" while providing resilience against quantum attacks. The specific throughput, latency and node-count figures are not present in the supplied text (the evaluation section is truncated), so no quantitative benchmark values can be reported here.
- **Information-theoretic security in the consensus core:** Because inter-node authentication relies on QKD-generated symmetric keys rather than computational hardness assumptions, the protocol's message integrity and source authentication do not degrade against an adversary with a quantum computer. The paper contrasts this explicitly with PQC, which it notes brings larger key sizes and reduced computational efficiency.
- **Liveness under primary failure (Theorem 1):** The system is claimed to distinguish malicious from honest client requests, finalize consensus only on 2f+1 consistent judgments, and update the configuration table based on primary-node performance. If a primary fails during the NEW phase, the stated cost is a delay of ΔT before proposal rights transfer to the next healthy node; if proposals have already passed TRANSMIT, the COMMIT phase can still finalize via the T_v mechanism.
- **Repudiation detection:** If a source node denies having authenticated a message, a verification node broadcasts a dispute message {m, m_α, 1}; receiving N−f−1 consistent dispute messages locally causes the source node to be judged as having attempted authentication repudiation.
- **Quorum structure:** Consensus thresholds throughout are 2f+1 (reached by collecting 2f consistent messages plus the node's own), with the underlying fault-tolerance bound N ≥ 3f+1 inherited from PBFT. Clients accept results after f+1 consistent replies.
- **Gap identified in prior work:** The paper argues that existing QKD-based blockchains (Kiktenko et al., Sun et al., Singh et al., Takaoğlu et al.) largely modify signatures inside *static* consensus schemes, and that dynamic BFT schemes (Duan and Zhang, Tang et al. with DKG plus BLS aggregate signatures, Li et al.'s DynaNet) incur significant communication overhead during membership transitions and do not address quantum resistance.

## Methodology in Plain English

The authors start from PBFT's structure — a primary node proposes, replicas vote, and a supermajority commits — and change two things.

First, they replace PBFT's view-change machinery with a hashing trick borrowed from distributed cache design. Imagine a clock face with 2³² positions. Every validator gets several labelled positions on it. To pick who proposes the next block, you take the hash of the previous block, drop that point onto the clock face, and walk clockwise until you hit a validator's label. Because the previous block hash is unpredictable, the choice is effectively random and rotates naturally block by block. Membership changes are handled by everyone agreeing, through the normal voting thresholds, on a new version of the table that maps labels to validators; the new table only takes effect once that agreement is reached.

Second, they replace digital signatures between validators with symmetric-key tags. A QKD network pre-distributes a distinct shared key to every ordered pair of nodes and stores it in a local key pool. To authenticate a message, the sender hashes the message and then, separately for each recipient, computes a keyed hash of that digest using the pairwise key, bundling all of these tags together. Each recipient checks only its own tag. Because the underlying keys have information-theoretic security, forging a tag is not a matter of computational effort — there is no shortcut an adversary, quantum or otherwise, can exploit.

Client traffic is handled differently: since ordinary users cannot be assumed to have QKD links, their requests and replies use post-quantum signatures. The protocol then runs through REQUEST, NEW (primary proposes a batch of G requests with digests and decision results), TRANSMIT (followers verify and broadcast their own decisions), COMMIT (a Merkle tree of approved proposals is committed), and REPLY.

## Why This Matters

**Impact on research.** The paper sits at the intersection of two usually separate literatures — quantum-secure blockchain architectures and dynamic-membership BFT — and argues that neither has adequately addressed the other. It offers a concrete design pattern (hash-ring leader election plus pairwise QKD authentication) that other protocol designers can borrow, and it reframes membership churn as a primary-rotation problem rather than a view-change problem.

**Real-world applications:**

- Consortium or permissioned blockchains operated by organizations that already run QKD infrastructure, such as telecom operators and government network providers.
- Long-lived records that must remain trustworthy for decades — land registries, supply-chain provenance, or regulatory archives — where "harvest now, decrypt later" attacks on today's signatures are a genuine concern.
- Cross-institutional settlement or data-sharing networks among banks and utilities where participants join and leave over time and static validator sets are impractical.
- Critical-infrastructure and energy-grid ledger systems, where the same operators often control both the quantum network and the blockchain nodes.

**Industry relevance.** The authors are affiliated with CAS Quantum Network Co., Ltd. and Anhui CAS Quantum Network Co., Ltd., so the work is aimed squarely at commercializing QKD infrastructure. The design choice to keep PQC signatures at the client edge while reserving QKD for the consensus core is a pragmatic deployment story: it avoids requiring every end user to have quantum hardware, which is the main barrier to QKD-based blockchain adoption.

## Future Directions

- **Quantitative validation at scale.** The supplied text claims performance comparable to PBFT but does not report throughput, latency, or tested node counts. A full evaluation across larger validator sets and realistic WAN latencies — particularly the cost of computing N−1 authentication tags per broadcast message — is the obvious next step.
- **The N−1 tag bottleneck.** Each broadcast requires the sender to compute a separate keyed hash for every peer. How this scales, and whether aggregation or Toeplitz-based fast hashing closes the gap, is left open in the available content.
- **Key-pool depletion and refresh.** QKD key pools are finite and rate-limited by the quantum channel. The paper describes lifecycle management (identification, renewal, updates) but does not report how key consumption under continuous consensus traffic is sustained.
- **Trust assumptions in the QKD layer.** Non-adjacent nodes obtain end-to-end keys via trusted-node relaying, which reintroduces a trust anchor. Combining QDBFT with measurement-device-independent or twin-field QKD, and formally analyzing security when a relay is compromised, remains open. Formal treatment of the truncated security proof for Theorem 1 and the carousel mapping is also incomplete in the available text.

## Target Audience

Protocol researchers and graduate students working on Byzantine fault tolerance and dynamic validator sets; engineers building consortium or permissioned blockchains who need a quantum-migration path; and quantum-network operators and architects evaluating what applications can justify QKD infrastructure investment. Readers without background in PBFT or QKD will find the design intelligible but will need the cited background literature for the security arguments.

## Authors’ abstract

The security foundation of blockchain system relies primarily on classical cryptographic methods and consensus algorithms. However, the advent of quantum computing poses a significant threat to conventional public-key cryptosystems based on computational hardness assumptions. In particular, Shor's algorithm can efficiently solve discrete logarithm and integer factorization problems in polynomial time, thereby undermining the immutability and security guarantees of existing…

[The full abstract on arXiv](https://arxiv.org/abs/2602.11606)

[Read the original paper](https://arxiv.org/abs/2602.11606)

---

Canonical: https://crypto.world/paper/qdbft-a-dynamic-consensus-algorithm-for-quantum-secured-blockchain-2602.11606
