Challenges of PBFT-Inspired Consensus for Blockchain and Enhancements over Neo dBFT

Abstract

Consensus mechanisms are a core feature for handling negotiation and agreements. Blockchain technology has seen the introduction of different sorts of consensus mechanism, ranging from tasks of heavy computation to the subtle mathematical proofs of Byzantine agreements. This paper presents the pioneer Delegated Byzantine Fault Tolerance (dBFT) protocol of Neo Blockchain, which was inspired by the Practical Byzantine Fault Tolerance (PBFT). Besides introducing its history, this study describes proofs and didactic examples, as well as novel design and extensions for Neo dBFT with multiple block proposals. Finally, we discuss challenges when dealing with strong Byzantine adversaries, and propose solutions inspired on PBFT for current weak-synchrony problems and increasing system robustness against attacks.

NeoCompiler Eco: experimentação de consenso em blockchain e contratos inteligentes

“…Tambémé possível executar a rede com umúnico nó de consenso, sem nenhum outro na camada de redistribuição peerto-peer, tornando a rede completamente suscetível a falhas bizantinas, porém reduzindo a sobrecarga de rede dos demais nós de consenso. Essa ferramenta tem possibilitado o desenvolvimento de trabalhos de cunho acadêmico na temática de consensos inovadores para blockchain, como [Coelho et al 2020, Araujo et al 2019].…”

A MILP Model for a Byzantine Fault Tolerant Blockchain Consensus

“…Each view increases the timeout by an exponential factor, so that it becomes too costly for an attacker to delay a message over a vTery long time (with delays growing over a linear factor). This is the same assumption from PBFT [11], and it is even more challenging to perform such attacks on a dynamic P2P network, as it would require controlling several independent nodes of the P2P infrastructure simultaneously, while the location of consensus nodes on dBFT is not made public [15].…”

“…Payloads can be received only once per view (Recv + is assigned to PrepReq + , PrepResp + , Commit + and ChView + ), which is defined with constraints created following the logic of Equation (15).…”

“…We considered the description of the dBFT 1.0 and 2.0 provided by Neo White Paper [10], Neo Project C# source-code (Neo dBFT source-code on GitHub: https://github.com/neo-project/neo), NeoResearch Community Yellow Paper (Neo dBFT source-code on GitHub: https://github.com/ neoresearch/yellowpaper), and recent blockchain discussions/enhancements for dBFT [15].…”

“…The limitations of the approach are also discussed, due to the time discretization techniques employed and high number of variables in the model, so efficient solvers and integer programming are needed. We refer the reader to the paper by Coelho et al [15] for more details on the dBFT paradigm and recent discussions on novel consensus algorithms for blockchain. To the best of our knowledge, this is the first time MILP is used to generate exact adversarial models for the verification of Byzantine consensus systems.…”