papers/zoo-fhe/zoo-fhe.tex

\documentclass[11pt,letterpaper]{article}
\usepackage{shared/zoocover}

% Packages
\usepackage[utf8]{inputenc}
% geometry loaded by zoocover
\usepackage{amsmath,amssymb,amsthm}
\usepackage{graphicx}
\usepackage{hyperref}
\usepackage{cite}
\usepackage{booktabs}
\usepackage{array}
\usepackage{tikz}
\usepackage{pgfplots}
\usepackage{algorithm}
\usepackage{algorithmic}
\usepackage{enumitem}
\usepackage{mathtools}
\usepackage{xcolor}

\pgfplotsset{compat=1.18}

% Theorem environments
\newtheorem{theorem}{Theorem}
\newtheorem{lemma}[theorem]{Lemma}
\newtheorem{proposition}[theorem]{Proposition}
\newtheorem{corollary}[theorem]{Corollary}
\theoremstyle{definition}
\newtheorem{definition}{Definition}
\newtheorem{example}{Example}
\theoremstyle{remark}
\newtheorem{remark}{Remark}

% Custom commands
\newcommand{\Zoo}{\textsc{Zoo}}
\newcommand{\Hanzo}{\textsc{Hanzo}}
\newcommand{\Lux}{\textsc{Lux}}
\newcommand{\TChain}{\textsc{T-Chain}}
\newcommand{\KChain}{\textsc{K-Chain}}
\newcommand{\TFHE}{\textsc{TFHE}}
\newcommand{\CKKS}{\textsc{CKKS}}
\newcommand{\MLKEM}{\textsc{ML-KEM}}
\newcommand{\NTT}{\textsc{NTT}}
\newcommand{\LWE}{\textsc{LWE}}
\newcommand{\RLWE}{\textsc{RLWE}}

% Title information
\title{\textbf{Threshold Fully Homomorphic Encryption on Zoo Network:\\GPU-Accelerated Confidential Computation}\\
\large Privacy-Preserving Smart Contracts via TFHE and CKKS}
\author{
Antje Worring, Zoo Labs Foundation Research\thanks{[email protected]}\\
\textit{Zoo Labs Foundation}\\
\texttt{[email protected]}
}

\date{September 2025}

\begin{document}

\zoocoverpage

\begin{abstract}
We present \Zoo{} Network's integration of threshold fully homomorphic encryption (\TFHE{} and \CKKS{}) into a Layer~2 blockchain, enabling arbitrary computation on encrypted on-chain state. \Zoo{} deploys programmable bootstrapping (\TFHE{}) for boolean and integer circuits and approximate arithmetic (\CKKS{}) for privacy-preserving machine-learning inference directly within smart contracts. By mapping the Number Theoretic Transform butterfly network onto GPU thread blocks, we achieve a \textbf{40$\times$ speedup} over CPU-only bootstrapping, reducing single-gate bootstrap latency from 12\,ms to 0.3\,ms. Ciphertext operations are distributed across validators via an Encrypt-to-Shares (E2S) threshold decryption protocol with a \textbf{67-of-100} default quorum, ensuring that no individual validator ever possesses a complete decryption key. Key shares are protected in transit with \MLKEM{}-768 post-quantum key encapsulation and anchored on \Lux{} \KChain{}. Formal correctness of the \CKKS{} rescaling pipeline, \TFHE{} noise budget tracking, and GPU scaling bounds are verified in Lean~4 (proofs reside in \texttt{Crypto/FHE/CKKS.lean}, \texttt{Crypto/FHE/TFHE.lean}, and \texttt{GPU/FHEScaling.lean}). We benchmark confidential DeFi order matching, sealed-bid auctions, and encrypted neural-network inference, demonstrating practical throughput for production deployment on \Zoo{} \TChain{}.
\end{abstract}

\tableofcontents
\newpage

%----------------------------------------------------------------------

\input{sections/01-introduction}
\input{sections/02-tfhe-on-zoo}
\input{sections/03-ckks-on-zoo}
\input{sections/04-gpu-acceleration}
\input{sections/05-threshold-decryption}
\input{sections/06-integration-with-t-chain}
\input{sections/07-formal-verification}
\input{sections/08-applications}
\input{sections/09-benchmarks}
\input{sections/10-related-work}
\input{sections/11-conclusion}
\input{sections/99-acknowledgments}

\end{document}