Back to Projects

Stochastic model of consensus in Ethereum Proof-of-Stake

Project Goals

The aim of this project is to create a streamlined modelling framework for Ethereum Proof-of-Stake Consensus with corresponding source code. This framework, using an agent-based approach, prioritises extensibility and computational efficiency. Unlike a one-to-one Ethereum agent-based simulation, our approach avoids excessive computational demands and data volume, focusing on essential consensus mechanisms. 


The research focuses on evaluating the resilience of the Ethereum Proof-of-Stake consensus protocol through simulations executed with the agent-based model. Firstly, we assess consensus efficiency under varying initial conditions, reflecting parameters like slot block gossiping latency and validator numbers, in a fully “honest” environment where validators willingly adhere to Ethereum consensus best practices. Validator heterogeneity is accounted for by modifying network topology and peer characteristics.

Additionally, we plan to introduce adversarial attacks on the consensus protocol to identify stress thresholds in a controlled environment. Addressing these challenges requires both theoretical and practical considerations. Theoretical aspects involve creating a minimalist model that encompasses fundamental protocol specifics, while the practical side entails translating this model into clear and functional open-source scripts for ease of replication and customisation by other users.