Master Thesis
Looking for Narwhals in the Asymmetric Sea
Consensus protocols are at the core of blockchains and cryptocurrency networks; a lot of innovation has happened in the space during the recent years.
DAG-Rider is a modular, elegant, and simple DAG-based consensus protocol proposed by Keidar et al. [1]. It builds upon reliable broadcast, common coin and gather to achieve Byzantine atomic broadcast. However, its use of reliable broadcast, which requires 3 communication rounds, negatively impacts its latency. Additionally, to ensure fairness, DAG-Rider requires unbounded memory, making it impractical for production systems. Danezis et al. [2] introduced the Narwhal and Tusk protocols to address these issues. Narwhal is a DAG-based mempool protocol that replaces reliable broadcast, while Tusk is a simple consensus protocol that operates on top of the DAG created by Narwhal. Other consensus protocols, such as Bullshark [4] and Shoal [5], also incorporate Narwhal in their communication layers.
Asymmetric distributed trust, introduced by Damgård et al. [6] and further developed by Alpos et al. [3], refers to trust configurations where participants have differing views of the trustworthiness of other nodes. Unlike traditional symmetric models, where trust is uniformly and globally shared, asymmetric trust enables nodes to individually specify their trust relationships, leading to more flexible but complex trust configurations. Alpos et al. [3] also introduced protocols to solve problems like reliable broadcast and binary consensus in this environment. More recently, Villacis et al. [7], introduced the first DAG-based consensus protocol for asymmetric settings by adapting DAG-Rider to use asymmetric quorums.
In this project, we will explore the Narwhal protocol, adapt it to use asymmetric quorums, prove its correctness, and possibly implement it.
References
[2] Narwhal and Tusk: A DAG-based Mempool and Efficient BFT Consensus
[3] Asymmetric Distributed Trust
[4] Bullshark: DAG BFT Protocols Made Practical
[5] Shoal: Improving DAG-BFT Latency And Robustness
[6] Secure Protocols with Asymmetric Trust
[7] An Asymmetric DAG-based Consensus Protocol, in preparation.