pith. sign in

arxiv: 2604.16053 · v1 · submitted 2026-04-17 · 💻 cs.DC

T-RBFT: A Scalable and Efficient Byzantine Consensus Based on Trusted Execution Environment for Consortium Blockchain

Wen Gao , Xinhong Hei , Yichuan Wang This is my paper

Pith reviewed 2026-05-10 07:54 UTC · model grok-4.3

classification 💻 cs.DC
keywords Byzantine consensusconsortium blockchaintrusted execution environmenttwo-layer consensusRaft protocolscalabilitypermissioned networkscommunication overhead
0
0 comments X

The pith

T-RBFT dynamically groups nodes and layers TEE-assisted BFT over Raft to cut consensus overhead in consortium blockchains.

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The paper proposes T-RBFT, a two-layer consensus mechanism for consortium blockchains that restricts participation to trusted entities. Nodes are first partitioned dynamically by runtime characteristics, after which an improved Raft protocol settles agreement inside each group and a TEE-assisted BFT protocol settles agreement between groups. Conventional BFT protocols exchange too many messages even when faults stay rare, so the design replaces that cost with lighter intra-group steps protected by hardware enclaves. A sympathetic reader would care because the result promises higher throughput and lower latency while preserving Byzantine safety in permissioned networks.

Core claim

T-RBFT first partitions consensus nodes into groups according to their observed runtime characteristics. Inside each group an enhanced Raft protocol reaches agreement efficiently under the assumption that most replicas remain honest. Between groups a BFT protocol assisted by trusted execution environments reaches agreement with far fewer messages. Experimental evaluation records lower communication volume, reduced end-to-end latency, and higher transaction throughput than prior two-layer protocols.

What carries the argument

The two-layer architecture of runtime-based dynamic grouping, Raft-based intra-group consensus, and TEE-assisted BFT for inter-group agreement.

If this is right

  • Total messages exchanged per consensus round drop because Raft handles most decisions locally.
  • Throughput rises as the system scales to more nodes while faults remain bounded.
  • End-to-end latency shrinks compared with full-mesh BFT under the same fault bound.
  • Permissioned networks gain a practical path to larger replica sets without proportional communication cost.

Where Pith is reading between the lines

These are editorial extensions of the paper, not claims the author makes directly.

  • The same grouping-plus-TEE pattern could be tested in other permissioned distributed systems that tolerate only a small fraction of faults.
  • Hardware dependence on TEE support may limit deployment to servers equipped with secure enclaves.
  • Frequent re-grouping could create transient performance dips if runtime characteristics fluctuate rapidly.

Load-bearing premise

Nodes can be grouped reliably by runtime characteristics without introducing new security risks or overhead that erase the efficiency gains.

What would settle it

A test run in which dynamic grouping repeatedly places faulty nodes together or TEE protections are bypassed, after which measured throughput fails to exceed that of standard two-layer BFT.

read the original abstract

With the continuous expansion of blockchain application scenarios, consortium chains have raised higher performance and security requirements for consensus mechanisms. Unlike public blockchains, consortium chains typically implement an admission mechanism that restricts participation to trusted entities, ensuring that most replicas are honest and the number of faulty nodes remains small under normal circumstances. In such settings, conventional Byzantine Fault Tolerant (BFT) protocols, which are designed for worst-case adversarial scenarios, incur excessive message exchanges and computational overhead, thereby limiting performance and scalability. To address this issue, this paper proposes T-RBFT, a two-layer consensus mechanism inspired by network sharding and enhanced by the trusted execution environment (TEE). In T-RBFT, consensus nodes are first dynamically grouped based on their runtime characteristics. Then, inter-group consensus is achieved through a TEE-assisted BFT protocol, while each group internally reaches agreement using an improved Raft-based mechanism. Experimental evaluation shows that T-RBFT reduces communication overhead and latency, and achieves higher throughput compared to existing two-layer consensus protocols, providing a scalable and communication-efficient consensus protocol for permissioned blockchain networks.

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, simulated authors' rebuttal, and a circularity audit. Tearing a paper down is the easy half of reading it; the pith above is the substance, this is the friction.

Referee Report

3 major / 2 minor

Summary. The paper proposes T-RBFT, a two-layer consensus protocol for consortium blockchains. Nodes are dynamically grouped according to runtime characteristics; intra-group agreement uses an improved Raft protocol (crash-fault tolerant), while inter-group consensus is performed by a TEE-assisted BFT protocol. The central claim is that this design reduces communication overhead and latency while delivering higher throughput than prior two-layer protocols, yielding a scalable and communication-efficient solution for permissioned networks.

Significance. If the security model and performance claims hold, the hybrid use of TEE for inter-group coordination combined with lightweight intra-group agreement could meaningfully improve efficiency in consortium settings where the fraction of faults is small. The approach builds on sharding ideas and TEE primitives in a way that may be of interest to the distributed-systems and blockchain communities, provided the fault-tolerance guarantees are rigorously established.

major comments (3)
  1. [§3] §3 (Protocol Design), intra-group layer description: The manuscript states that each dynamically formed group reaches agreement via an improved Raft protocol. Raft (even when improved) tolerates only crash faults and fails in the presence of a single Byzantine node that sends conflicting messages or withholds votes. No mechanism—such as TEE-based attestation, runtime behavioral filtering, or cryptographic admission control—is described that would prevent a Byzantine node from mimicking honest runtime characteristics long enough to enter a group. Once inside, that node can stall or corrupt the group's local consensus; the inter-group TEE-BFT layer operates only on group outputs and cannot repair internal corruption. This assumption is load-bearing for the claimed Byzantine tolerance and performance gains.
  2. [§5] §5 (Experimental Evaluation): The abstract and conclusion assert that T-RBFT reduces communication overhead, lowers latency, and achieves higher throughput than existing two-layer protocols. However, the manuscript supplies no details on experimental setup (hardware, network topology, latency/bandwidth parameters), baseline implementations, number of runs, statistical methods, or raw data. Without these, the performance claims cannot be verified or reproduced, undermining the central empirical support for the protocol's practicality.
  3. [§3.3] §3.3 (Inter-group TEE-BFT): The TEE-assisted BFT layer is presented as delivering net security and efficiency gains. The manuscript does not specify the precise TEE threat model (e.g., side-channel attacks, rollback, or compromised enclave code) nor quantify the additional overhead or new attack surface introduced by TEE usage relative to a pure software BFT protocol.
minor comments (2)
  1. [§3.1] Notation for group formation and runtime characteristics is introduced without a formal definition or pseudocode; a concise algorithm box would improve clarity.
  2. [Abstract, §1] Several sentences in the abstract and introduction repeat the same performance claims; tightening the wording would reduce redundancy.

Simulated Author's Rebuttal

3 responses · 0 unresolved

We thank the referee for the thoughtful and constructive report. The comments identify important areas for clarification and strengthening, particularly regarding security assumptions, experimental reproducibility, and the TEE threat model. We address each major comment below and indicate the revisions we will make in the next version of the manuscript.

read point-by-point responses

  1. Referee: [§3] §3 (Protocol Design), intra-group layer description: The manuscript states that each dynamically formed group reaches agreement via an improved Raft protocol. Raft (even when improved) tolerates only crash faults and fails in the presence of a single Byzantine node that sends conflicting messages or withholds votes. No mechanism—such as TEE-based attestation, runtime behavioral filtering, or cryptographic admission control—is described that would prevent a Byzantine node from mimicking honest runtime characteristics long enough to enter a group. Once inside, that node can stall or corrupt the group's local consensus; the inter-group TEE-BFT layer operates only on group outputs and cannot repair internal corruption. This assumption is load-bearing for the claimed Byzantine tolerance and performance gains.

    Authors: We agree that the intra-group Raft layer provides only crash-fault tolerance and that a single Byzantine node inside a group could disrupt local consensus. The manuscript's design premise, stated in the abstract and introduction, is that consortium blockchains use an admission mechanism restricting participation to trusted entities, so that the fraction of Byzantine nodes remains small. Dynamic grouping occurs only among these already-admitted nodes. We acknowledge that the current text does not explicitly describe additional runtime checks or attestation at the grouping stage beyond the consortium admission control. In the revision we will add a clarifying paragraph in §3 stating the reliance on out-of-band admission control, note the assumption that admitted nodes do not behave as Byzantine during grouping, and discuss the residual risk if that assumption is violated. We will also emphasize that the inter-group TEE-BFT layer aggregates group outputs and provides Byzantine tolerance at the system level under the low-fault-fraction regime. revision: partial

  2. Referee: [§5] §5 (Experimental Evaluation): The abstract and conclusion assert that T-RBFT reduces communication overhead, lowers latency, and achieves higher throughput than existing two-layer protocols. However, the manuscript supplies no details on experimental setup (hardware, network topology, latency/bandwidth parameters), baseline implementations, number of runs, statistical methods, or raw data. Without these, the performance claims cannot be verified or reproduced, undermining the central empirical support for the protocol's practicality.

    Authors: We concur that the experimental section is insufficiently detailed for reproducibility. The current manuscript reports only high-level performance trends without hardware specifications, network parameters, baseline protocol details, run counts, or statistical methods. In the revised version we will expand §5 with: (1) hardware and VM configurations, (2) network topology, latency, and bandwidth settings, (3) exact baseline two-layer protocols and their implementations, (4) number of independent runs and warm-up periods, (5) statistical reporting (means, standard deviations, confidence intervals), and (6) a note on availability of experimental scripts or data. These additions will allow verification of the claimed reductions in overhead and improvements in latency/throughput. revision: yes

  3. Referee: [§3.3] §3.3 (Inter-group TEE-BFT): The TEE-assisted BFT layer is presented as delivering net security and efficiency gains. The manuscript does not specify the precise TEE threat model (e.g., side-channel attacks, rollback, or compromised enclave code) nor quantify the additional overhead or new attack surface introduced by TEE usage relative to a pure software BFT protocol.

    Authors: We thank the referee for noting the missing TEE security analysis. The manuscript assumes a standard TEE model (enclave integrity and confidentiality) but does not enumerate side-channel, rollback, or enclave-compromise threats, nor compare overhead or attack surface to pure-software BFT. In the revision we will insert a new paragraph in §3.3 that: (a) states the assumed TEE threat model (e.g., no side-channel leakage under current mitigations, no rollback via secure storage), (b) references common TEE literature for these assumptions, (c) qualitatively quantifies enclave transition and attestation overhead relative to software BFT, and (d) argues why the hybrid design still yields net efficiency gains in the low-fault consortium setting despite the added TEE surface. This will make the security claims more rigorous. revision: yes

Circularity Check

0 steps flagged

No circularity: protocol design and experimental claims are independent of inputs

full rationale

The paper presents T-RBFT as a new two-layer protocol design (dynamic grouping by runtime characteristics, TEE-assisted inter-group BFT, improved Raft intra-group) whose performance claims rest on explicit experimental comparisons to prior two-layer protocols. No equations, fitted parameters, or predictions are defined in terms of themselves; the derivation chain consists of the protocol specification and benchmark results, which are externally falsifiable and do not reduce to self-citation or redefinition. Self-citations, if present, are not load-bearing for the central claims.

Axiom & Free-Parameter Ledger

0 free parameters · 1 axioms · 0 invented entities

The design rests on the domain assumption that consortium settings have admission control keeping faulty nodes few, plus the implicit assumption that TEE hardware is available and secure enough to provide net benefit.

axioms (1)
  • domain assumption Consortium chains implement admission mechanisms that restrict participation to trusted entities so that most replicas are honest and the number of faulty nodes remains small under normal circumstances.
    This premise is stated in the abstract as the reason conventional BFT overhead is excessive.

pith-pipeline@v0.9.0 · 5499 in / 1272 out tokens · 47819 ms · 2026-05-10T07:54:34.145662+00:00 · methodology

discussion (0)

Sign in with ORCID, Apple, or X to comment. Anyone can read and Pith papers without signing in.

Reference graph

Works this paper leans on

36 extracted references · 36 canonical work pages

  1. [1]

    Information Processing & Management58(1), 102436 (2021)

    Xu, X., Sun, G., Luo, L., Cao, H., Yu, H., Vasilakos, A.V.: Latency performance modeling and analysis for hyperledger fabric blockchain network. Information Processing & Management58(1), 102436 (2021)

    work page 2021

  2. [2]

    ACM Transactions on Computer Systems (TOCS)20(4), 398–461 (2002)

    Castro, M., Liskov, B.: Practical byzantine fault tolerance and proactive recovery. ACM Transactions on Computer Systems (TOCS)20(4), 398–461 (2002)

    work page 2002

  3. [3]

    Peer-to-Peer Networking and Applications14(5), 2826– 2839 (2021)

    Li, Y., Qiao, L., Lv, Z.: An optimized byzantine fault tolerance algorithm for consortium blockchain. Peer-to-Peer Networking and Applications14(5), 2826– 2839 (2021)

    work page 2021

  4. [4]

    In: 2021 IEEE 12th International Conference on Software Engineering and Service Science (ICSESS), pp

    Li, W., He, M.: Ebft: A hierarchical and group-based byzantine fault tolerant consensus algorithm. In: 2021 IEEE 12th International Conference on Software Engineering and Service Science (ICSESS), pp. 32–37 (2021). IEEE

    work page 2021

  5. [5]

    In: Proceedings of the 2021 ACM Symposium on Principles of Distributed Computing, pp

    Fitzi, M., Liu-Zhang, C.-D., Loss, J.: A new way to achieve round-efficient byzan- tine agreement. In: Proceedings of the 2021 ACM Symposium on Principles of Distributed Computing, pp. 355–362 (2021)

    work page 2021

  6. [6]

    Mathematics8(10), 1673 (2020) 22

    Choi, S.-M., Park, J., Jang, K., Park, C.: Rapid consensus structure: Continu- ous common knowledge in asynchronous distributed systems. Mathematics8(10), 1673 (2020) 22

    work page 2020

  7. [7]

    Future Generation Computer Systems124, 33–48 (2021)

    Fan, Y., Wu, H., Paik, H.-Y.: Dr-bft: A consensus algorithm for blockchain-based multi-layer data integrity framework in dynamic edge computing system. Future Generation Computer Systems124, 33–48 (2021)

    work page 2021

  8. [8]

    Science China Information Sciences64(2), 121101 (2021)

    Fu, X., Wang, H., Shi, P.: A survey of blockchain consensus algorithms: mecha- nism, design and applications. Science China Information Sciences64(2), 121101 (2021)

    work page 2021

  9. [9]

    IEEE Transactions on Systems, Man, and Cybernetics: Systems51(4), 2415–2424 (2019)

    Xiang, F., Huaimin, W., Peichang, S.: Proof of previous transactions (popt): An efficient approach to consensus for jcledger. IEEE Transactions on Systems, Man, and Cybernetics: Systems51(4), 2415–2424 (2019)

    work page 2019

  10. [10]

    Wang, H., Guo, K., Pan, Q.: Byzantine fault tolerance consensus algorithm based on voting mechanism. J. Comput. Appl39(06), 1766–1771 (2019)

    work page 2019

  11. [11]

    Computer Networks222, 109541 (2023)

    Tian, J., Tian, J., Xu, H.: Tsbft: A scalable and efficient leaderless byzantine consensus for consortium blockchain. Computer Networks222, 109541 (2023)

    work page 2023

  12. [12]

    IEEE Transactions on Information Forensics and Security18, 2276–2291 (2023)

    Liu, Y., Xing, X., Cheng, H., Li, D., Guan, Z., Liu, J., Wu, Q.: A flexible shard- ing blockchain protocol based on cross-shard byzantine fault tolerance. IEEE Transactions on Information Forensics and Security18, 2276–2291 (2023)

    work page 2023

  13. [13]

    IEEE Transactions on Dependable and Secure Computing 21(4), 2478–2493 (2023)

    Jalalzai, M.M., Niu, J., Feng, C., Gai, F.: Fast-hotstuff: A fast and robust bft pro- tocol for blockchains. IEEE Transactions on Dependable and Secure Computing 21(4), 2478–2493 (2023)

    work page 2023

  14. [14]

    In: Proceedings of the 2019 ACM Symposium on Principles of Distributed Computing, pp

    Yin, M., Malkhi, D., Reiter, M.K., Gueta, G.G., Abraham, I.: Hotstuff: Bft consensus with linearity and responsiveness. In: Proceedings of the 2019 ACM Symposium on Principles of Distributed Computing, pp. 347–356 (2019)

    work page 2019

  15. [15]

    Peer-to-Peer Networking and Applications16(1), 420–430 (2023)

    Qi, J., Guan, Y.: Practical byzantine fault tolerance consensus based on com- prehensive reputation. Peer-to-Peer Networking and Applications16(1), 420–430 (2023)

    work page 2023

  16. [16]

    Information Processing & Management59(2), 102884 (2022)

    Chen, Y., Li, M., Zhu, X., Fang, K., Ren, Q., Guo, T., Chen, X., Li, C., Zou, Z., Deng, Y.: An improved algorithm for practical byzantine fault tolerance to large- scale consortium chain. Information Processing & Management59(2), 102884 (2022)

    work page 2022

  17. [17]

    IEEE Transactions on Computers62(1), 16–30 (2011)

    Veronese, G.S., Correia, M., Bessani, A.N., Lung, L.C., Verissimo, P.: Efficient byzantine fault-tolerance. IEEE Transactions on Computers62(1), 16–30 (2011)

    work page 2011

  18. [18]

    IEEE Transactions on Computers68(1), 139– 151 (2018)

    Liu, J., Li, W., Karame, G.O., Asokan, N.: Scalable byzantine consensus via hardware-assisted secret sharing. IEEE Transactions on Computers68(1), 139– 151 (2018)

    work page 2018

  19. [19]

    IEEE Transactions on Dependable and Secure Computing19(6), 3621–3639 (2021)

    Feng, X., Ma, J., Miao, Y., Liu, X., Choo, K.-K.R.: Social characteristic-based 23 propagation-efficient pbft protocol to broadcast in unstructured overlay networks. IEEE Transactions on Dependable and Secure Computing19(6), 3621–3639 (2021)

    work page 2021

  20. [20]

    Computer Communications231, 108032 (2025)

    Lu, L., Sun, L., Zou, Y.: An efficient sharding consensus protocol for improving blockchain scalability. Computer Communications231, 108032 (2025)

    work page 2025

  21. [21]

    Concurrency and Computation: Practice and Experience34(10), 6813 (2022)

    Singh, J., Kumawat, A., Venkatesan, S.: Improved byzantine fault tolerance with fast consensus. Concurrency and Computation: Practice and Experience34(10), 6813 (2022)

    work page 2022

  22. [22]

    IEEE Internet of Things Journal8(6), 4291–4304 (2020)

    Huang, C., Wang, Z., Chen, H., Hu, Q., Zhang, Q., Wang, W., Guan, X.: Repchain: A reputation-based secure, fast, and high incentive blockchain system via sharding. IEEE Internet of Things Journal8(6), 4291–4304 (2020)

    work page 2020

  23. [23]

    The Journal of Supercomputing78(4), 4850–4881 (2022)

    Li, C., Zhang, J., Yang, X.: Scalable blockchain storage mechanism based on two-layer structure and improved distributed consensus. The Journal of Supercomputing78(4), 4850–4881 (2022)

    work page 2022

  24. [24]

    IEEE Transactions on Vehicular Technology73(9), 13828–13838 (2024)

    Zhao, F., Yang, B., Li, C., Zhang, C., Zhu, L., Liang, G.: Two-layer consen- sus based on primary-secondary consortium chain data sharing for internet of vehicles. IEEE Transactions on Vehicular Technology73(9), 13828–13838 (2024)

    work page 2024

  25. [25]

    IET Blockchain4, 555–569 (2024)

    Yuan, H., Li, F., Diao, R., Shu, T.: Double-layer byzantine fault-tolerant grouping consensus algorithm based on raft. IET Blockchain4, 555–569 (2024)

    work page 2024

  26. [26]

    Complex & Intelligent Systems9(2), 1507–1524 (2023)

    Xie, M., Liu, J., Chen, S., Xu, G., Lin, M.: Primary node election based on probabilistic linguistic term set with confidence interval in the pbft consensus mechanism for blockchain. Complex & Intelligent Systems9(2), 1507–1524 (2023)

    work page 2023

  27. [27]

    In: 2014 USENIX Annual Technical Conference (USENIX ATC 14), pp

    Ongaro, D., Ousterhout, J.: In search of an understandable consensus algorithm. In: 2014 USENIX Annual Technical Conference (USENIX ATC 14), pp. 305–319 (2014)

    work page 2014

  28. [28]

    Peer-to-Peer Networking and Applications15(2), 1008–1028 (2022)

    Cao, X., Zhang, J., Wu, X., Liu, B.: A survey on security in consensus and smart contracts. Peer-to-Peer Networking and Applications15(2), 1008–1028 (2022)

    work page 2022

  29. [29]

    Chinese Journal of Electronics33(3), 623–634 (2024)

    Wang, Y., Gao, W., Hei, X., Du, Y.: Method and practice of trusted embed- ded computing and data transmission protection architecture based on android. Chinese Journal of Electronics33(3), 623–634 (2024)

    work page 2024

  30. [30]

    In: 2023 7th International Conference on Computing Methodologies and Communication (ICCMC), pp

    Kamal, K.K., Gupta, S., Joshi, P., Kapoor, M.: Secure mobile id architecture on android devices based on trust zone. In: 2023 7th International Conference on Computing Methodologies and Communication (ICCMC), pp. 1060–1064 (2023). IEEE

    work page 2023

  31. [31]

    In: 2022 International Conference on Networking and Network Applications (NaNA), pp

    Zhang, T., Cui, X., Wang, Y., Du, Y., Gao, W.: Tcs security analysis in intel 24 sgx enclave multithreading. In: 2022 International Conference on Networking and Network Applications (NaNA), pp. 276–281 (2022). IEEE

    work page 2022

  32. [32]

    IACR Transactions on Cryptographic Hardware and Embedded Systems2024(2), 714–734 (2024)

    Svenda, P., Dufka, A., Broz, M., Lacko, R., Jaros, T., Zatovic, D., Pospisil, J.: Tpmscan: A wide-scale study of security-relevant properties of tpm 2.0 chips. IACR Transactions on Cryptographic Hardware and Embedded Systems2024(2), 714–734 (2024)

    work page 2024

  33. [33]

    Peer-to-Peer Networking and Applications 16(3), 1327–1339 (2023)

    Wang, Y., Wang, W., Zeng, Y., Yang, T.: Gradingshard: A new sharding protocol to improve blockchain throughput. Peer-to-Peer Networking and Applications 16(3), 1327–1339 (2023)

    work page 2023

  34. [34]

    In: Proceedings of the 2021 International Conference on Management of Data, pp

    Amiri, M.J., Agrawal, D., El Abbadi, A.: Sharper: Sharding permissioned blockchains over network clusters. In: Proceedings of the 2021 International Conference on Management of Data, pp. 76–88 (2021)

    work page 2021

  35. [35]

    In: Proceedings of the 23rd ACM/IFIP International Middleware Conference, pp

    Messadi, I., Becker, M.H., Bleeke, K., Jehl, L., Mokhtar, S.B., Kapitza, R.: Splitbft: Improving byzantine fault tolerance safety using trusted compartments. In: Proceedings of the 23rd ACM/IFIP International Middleware Conference, pp. 56–68 (2022)

    work page 2022

  36. [36]

    Computer Communications201, 102–115 (2023) 25

    Yadav, A.K., Singh, K., Amin, A.H., Almutairi, L., Alsenani, T.R., Ahmadian, A.: A comparative study on consensus mechanism with security threats and future scopes: Blockchain. Computer Communications201, 102–115 (2023) 25

    work page 2023