Performance modeling, analysis, and tuning of blockchain networks

Abstract

Blockchain performance modeling can be used to provide us with a deeper understanding of the behaviour and dynamics within blockchain peer-to-peer networks. A theoretical model will help the blockchain designers obtain a better understanding of the underlying blockchain dynamics and characteristics which impact the performance of the blockchain network. A theoretical model can also accelerate the development of a blockchain system by quickly discovering a theoretically optimal initial design, which is preferable to an incremental design through iterative benchmarking. In this research, we develop theoretical models for both public and private blockchains. We use these models to investigate the impact of the network topology on three different aspects of blockchain networks in terms of performance, security (forks) and decentralization. Since the consensus mechanism in the majority of public blockchains is proof-of-work (PoW), we first develop a performance model for the original and well-established Bitcoin blockchain network. The proposed model can be easily adapted for other well-known PoW-based blockchains such as Ethereum. Blockchain networks that employ PoW in their consensus mechanism may face inconsistencies in the form of forks. In this research, we investigate the cause and length of forks for the Bitcoin network. In addition, we present a methodology for quantifying the decentralization degree of a blockchain network. To accomplish these objectives, we use two well-known graph models of Erdös-Rény and Barabási–Albert in order to study the blockchain network topology. We also adapt our model to study the impact of deploying a relay network and investigate the effect of the relay network size on the network performance and decentralization of PoW-based blockchains. For verifying and validating the developed models, we use historical data mined from the Bitcoin network as well as results obtained from simulation using OMNet++. On the other hand, Byzantine Fault-Tolerant (BFT) protocols are classical algorithms that offer a faster and more energy-efficient consensus mechanism compared to PoW. Hotstuff is a partially synchronous BFT State Machine Replication (SMR) protocol that aims to address the performance and scalability issues commonly found in PoW. We propose a framework for developing a performance model for BFT-based blockchains and particularly focus on HotStuff. Since HotStuff and other BFT-based blockchains have an almost similar mechanism, this model can be also adapted for other variants of BFT consensus.

Date	8 Dec 2022
Original language	American English
Awarding Institution	École de technologie supérieure
Supervisor	Kaiwen Zhang (Supervisor) & Chamseddine Talhi (Co-supervisor)