Unified Cryptographic Proof-of-Execution and Eco-Compute Optimization for Decentralized Autonomous AI Infrastructure

Le résumé fourni par la source

ABSTRACT As autonomous artificial intelligence (AI) agents and edge compute micro-nodes assume central roles in critical digital systems, guaranteeing computational fidelity without compromising hardware efficiency or user privacy remains an imperative challenge. Existing attestation models heavily rely on centralized Trusted Execution Environments (TEEs), which exhibit side-channel vulnerabilities and restricted memory bandwidth. This paper presents a comprehensive, multi-layered architecture uniting zero-knowledge Proof-of-Execution (ZK-PoE) with hardware-abstracted eco- compute pipelines. Developed under the research mandate of XR6 Labs (HQ) — Research & Engineering, our implementation leverages layer-wise Polynomial Commitments (KZG) and succinct non-interactive arguments (zk- SNARKs) to verify matrix multiplication traces in real time. Empirical benchmarks across heterogeneous micro-node clusters reveal an 87.4% reduction in proof generation latency and an 88.2% drop in per-inference energy consumption relative to baseline zero-knowledge virtual machines (zkVMs). These results confirm the feasibility of sustainable, verifiable, and ethical AI compute deployment globall

Ce résumé expose les affirmations des auteurs. BNTIC ne l’interprète pas comme une validation indépendante des résultats.