.png){kind=icon}
Return
Secure and efficient asynchronous BFT consensus leveraging trusted execution environments
.png){kind=icon}
PRE
AI
DOI：10.1016/j.future.2026.108786.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
Asynchronous Byzantine Fault Tolerance (ABFT) protocols enable distributed consensus without timing assumptions, but suffer from high cryptographic overhead, quadratic message complexity, and impractical deployment requirements such as universal trusted hardware or expensive threshold encryption. Existing TEE-assisted BFT protocols operate only under partial synchrony and assume all nodes possess trusted execution capabilities, limiting their applicability in heterogeneous environments. We present a new TEE-assisted asynchronous BFT protocol, TruBFT, that supports partial TEE deployment, thereby removing the requirement that all nodes be TEE-enabled. In our design, non-TEE nodes form temporary groups and delegate consensus tasks to selected TEE-equipped nodes. After reaching consensus, the result is returned and the group is dissolved, enabling lightweight, on-demand participation. To reduce the overhead of traditional threshold encryption, we leverage TEE-protected communication and enclave-confined cryptographic processing, together with asymmetric encryption, to efficiently resist censorship attacks. Additionally, we redesign the asynchronous binary agreement phase by executing all vote-related operations inside enclaves, ensuring confidentiality of ABA voting information and improving resistance to collusion. We formally analyze the protocol’s security under the asynchronous model with partially trusted hardware. Experimental results confirm its security and performance, making TruBFT suitable for real-world deployment in decentralized systems with heterogeneous trust assumptions. Experiments conducted across up to 16 virtual machines on Alibaba Cloud over wide area networks show that TruBFT maintains a stable throughput of approximately 476 tx/s under a load of 100000 transactions.
Journal
F
IF:
6.1
Papers:
6.8K
Citations:
2.3W
Organization
View more organizations
Cited Papers
No cited papers available