Files
zeekay abf99d05fc papers: reorganize into per-paper dirs, add Antje Worring as co-author
Restructured from flat .tex/.pdf files into per-paper subdirectories.
Added Antje Worring as co-author on all papers.
Synced PDFs to pdfs/ for download site.
2026-04-18 21:50:07 -07:00

42 lines
2.3 KiB
TeX

\section{Introduction}
\label{sec:introduction}
Distributed consensus---the problem of achieving agreement among a set of processes in
the presence of failures---is foundational to replicated state machines, distributed
databases, and blockchain systems. Classical Byzantine fault tolerant (BFT) protocols
such as PBFT~\cite{castro1999practical} and HotStuff~\cite{yin2019hotstuff} achieve
deterministic safety under $f < n/3$ Byzantine faults but exhibit $O(n^2)$ or $O(n)$
communication complexity per decision, limiting scalability.
Nakamoto consensus~\cite{nakamoto2008bitcoin} scales to thousands of participants but
achieves only probabilistic safety that strengthens slowly over time, requiring minutes
to hours for practical finality. The metastable protocol family---Slush, Flare, Wave,
and Lux---introduced a fundamentally different paradigm: consensus through repeated
stochastic sampling, achieving $O(\log n)$ convergence with $O(kn)$ message complexity.
This paper formalizes the Wave protocol, the third member of the metastable family,
which introduces \emph{confidence counters} that track the full history of supermajority
observations. Unlike Flare's conviction counter (which tracks only consecutive rounds),
Wave's confidence counters are cumulative and monotone---once a color gains a lead,
the adversary must overcome the entire confidence history to reverse the decision.
\paragraph{Contributions.}
\begin{enumerate}[nosep]
\item A complete formal specification of the Wave protocol with DAG-based transaction
ordering and confidence counters (\S\ref{sec:protocol}).
\item A proof that confidence counters exhibit strict monotonicity under honest majority:
once color $c$ leads by margin $\delta$, the probability that a competing color catches
up decays exponentially in $\delta$ (\S\ref{sec:analysis}).
\item A Markov chain model of counter evolution yielding closed-form convergence bounds
(\S\ref{sec:markov}).
\item A proof that strong preference in the DAG is a monotone predicate under Wave
dynamics (\S\ref{sec:dag}).
\item Empirical evaluation on a 500-node testnet demonstrating sub-second finality
(\S\ref{sec:evaluation}).
\end{enumerate}
%% -----------------------------------------------------------------------
%% 2. SYSTEM MODEL
%% -----------------------------------------------------------------------