Helena Carolina Delgado Teixeira

Co-advisor: M. Matos


Self-Adapting BFT Consensus: Leveraging Heterogeneity in Dissemination/Aggregation Trees


Tese submetida para provas de mestrado em Engenharia Informática e de Computadores Instituto Superior Técnico, Universidade de Lisboa.

Abstract

Permissioned blockchains are a class of blockchains where the processes that run consensus are lim- ited and known by all participants. These blockchains can execute variants of Byzantine consensus that offer finality. An approach to support a large number of participants in this context is to use dissemination and aggregation trees to support the communication required to execute protocol rounds. Previous work using this topology is either topology-agnostic or assumes homogeneous environments. Many Byzantine consensus protocols are leader-based, and in the blockchain scenario, there are com- pelling reasons to rotate the leader between consecutive consensus instances, such as the distribution of the load of the leader and censorship resistance. This work addresses the challenges of implement- ing a rotating leader policy in BFT consensus that uses dissemination and aggregation trees, and we propose topology-aware heuristics to create dissemination and aggregation trees in heterogeneous en- vironments. Through simulations, we evaluate the performance of our heuristics in realistic scenarios, showing that they can reduce the average time needed to collect a quorum by 70%.

Publicações

Self-Adapting BFT Consensus: Leveraging Heterogeneity in Dissemination/Aggregation Trees
Helena Carolina Delgado Teixeira
MSc Thesis. Instituto Superior Técnico, Universidade de Lisboa.
November 2023.
Available BibTeX, MSC Thesis, and extended abstract, and mid-term report.
Disseminação e Agregação Cientes da Topologia para Suportar Consenso Bizantino em Larga Escala.
H. Teixeira, L. Rodrigues and M. Matos.
Actas do décimo quarto Simpósio de Informática (Inforum), Porto, Portugal, Sep. 2023.
Available BibTeX, extended report (pdf).

Luís Rodrigues