Actor-based Model for Concurrent Byzantine Fault-tolerant Algorithm

Author:Chi Zhang, Rong Wang, Wei-Tek Tsai, Juan He, Can Liu, Qi Li

Affiliations:School of Beihang University,Beijing Tiande Technologies

Abstract:Consensus mechanism is the core element of blockchain (BC) technology. A good

consensus mechanism can improve the performance of BC system and promote the application of BC technology. The most widely used consensus algorithms are Proof of Work (PoW) algorithm and the Proof of Stake (PoS) algorithm. However, only 3 to 7 transactions can be processed per second, which means the BC system does not meet the performance requirements of commercial application. The Practical Byzantine Fault Tolerance (PBFT) algorithm provides (n-1)/3 fault-tolerance under the premise of ensuring liveness and safety, which can process hundreds of transactions per second. However, large commercial BC applications often require higher performance and the throughput of BC system need to reach thousands of transactions per second (TPS). To further improve system performance, we propose actor-based model for Concurrent Byzantine Fault Tolerant (CBFT) algorithm, which uses actor model to achieve higher concurrency and improve the transaction processing speed of BC system. We design, development and test a BC system, which is based on CBFT algorithm. The experimental results show that the algorithm can maximize the bandwidth of the network at the optimal block interval time of 3 seconds, the TPS can reach 1500-2000, and the delay can be controlled between 100ms and 1000ms, which can meet the performance requirements of most commercial systems.

Keywords:Blockchain, consensus mechanism, actor model, concurrent byzantine fault-tolerant algorithm.