mirror of
https://github.com/lux-apps/papers.git
synced 2026-10-30 18:15:21 +00:00
zeekay
abf99d05fc
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.
31 lines
1.2 KiB
TeX
31 lines
1.2 KiB
TeX
\section{Security Analysis}
|
|
|
|
\label{sec:security}
|
|
|
|
\subsection{Eclipse Attacks}
|
|
|
|
An eclipse attack attempts to isolate a validator's sample to consist only of Byzantine
|
|
nodes. With uniform random sampling from $n = 500$ validators and $k = 20$:
|
|
\[
|
|
\Pr[\text{all } k \text{ sampled are Byzantine}] = \binom{f}{k} / \binom{n}{k}
|
|
\leq \left(\frac{f}{n}\right)^k = (0.33)^{20} < 2^{-32}
|
|
\]
|
|
Network-level defenses (peer diversity, multiple transport protocols) further reduce this
|
|
probability in practice.
|
|
|
|
\subsection{Sybil Resistance}
|
|
|
|
The Wave protocol assumes a fixed, authenticated validator set. Sybil resistance is
|
|
provided by the staking mechanism on the platform chain: each validator must bond
|
|
stake, and the sampling probability is proportional to stake weight.
|
|
|
|
\subsection{Long-Range Attacks}
|
|
|
|
Wave's DAG structure and confidence counters are ephemeral---they concern only the
|
|
current UTXO set. Long-range attacks (rewriting history from a past state) are prevented
|
|
by the platform chain's periodic checkpointing.
|
|
|
|
%% -----------------------------------------------------------------------
|
|
%% 11. CONCLUSION
|
|
%% -----------------------------------------------------------------------
|