LgTTBFT : Effective Byzantine Fault Tolerance Algorithm Based on Structured Network and Trusted Execution Environment

EasyChair Preprint 1713

Abstract

The BFT consensus algorithm provides 100% security rather than probability, but it has not been widely used due to its high resource consumption and low consensus efficiency. The blockchain community has also been optimized on the basis of traditional Byzantine fault tolerance, trying to actually apply the BFT protocol to the consensus layer of the blockchain. In this paper, we propose Loop-Grouping Tree and TEE BFT (LgTTBFT), a simple and efficient Byzantine fault tolerant consensus strategy. At the heart of LgTTBFT is a new tree topology (LgTree) and Trusted Execution Environment (TEE). We constructed a structured network and multi-reconstructed multi-way tree structure LgTree as the node organization structure of the network layer of the blockchain system. Compared with MinBFT and Ethereum, it provides a simpler network structure and more efficient routing efficiency. The special tree structure also performs well in terms of load balancing and stability. The use of TEE has greatly improved system efficiency from both the number of replicas and the communication process: reduce the minimum number of replicas from 3f +1 to 2f +1, The communication phase of the consensus process is reduced from three phases of PBFT to two phases, and the communication complexity of the PREPARE phase is reduced from O(n2) to O(n). Experiments have shown that the combination of structured network and trusted execution environment provides better throughput and latency.

Keyphrases: Blockchain, Byzantine fault tolerance, Structured Network, state machine replication, trusted components

@booklet{EasyChair:1713,
  author    = {Rihong Wang and Na Li and Quanqing Xu and Lifeng Zhang and Congying Xing},
  title     = {LgTTBFT : Effective Byzantine Fault Tolerance Algorithm Based on Structured Network and Trusted Execution Environment},
  howpublished = {EasyChair Preprint 1713},
  year      = {EasyChair, 2019}}