Cellular Inference Mesh: Salabim DES of edge-cloud LLM inference under PACELC saturation

Fuente: Zenodo
Guardado en:
Detalles Bibliográficos
Autor principal: Flores, Carlos Ulisses
Formato: Recurso digital
Lenguaje:inglés
Publicado: Zenodo 2026
Materias:
Acceso en línea:
Etiquetas: Agregar Etiqueta
Sin Etiquetas, Sea el primero en etiquetar este registro!
_version_ 1866901616329752576
author Flores, Carlos Ulisses
author_facet Flores, Carlos Ulisses
contents Discrete-event simulation in Salabim of the Cellular Inference Mesh edge-cloud architecture for industrial multilingual AGV voice command processing under PACELC. Implements 300 Monte-Carlo replicates over 3 scenarios and 2 arms for 1800 simulated seconds, with parametric variation per replicate over realistic operational ranges. Verifies the Pareto-Efficiency Theorem (over 5 axes against 256 sub-compositions and 7 external architectures) and articulates the Saturated Pipeline Conjecture, which explains the structural insensitivity of percentile-99 latency to fallback-rate reductions under edge saturation. Reproducibility guaranteed via SHA-256 hash chain, seed_canonical=42, 53 pytest tests with bit-parity.
format Recurso digital
id zenodo_https___doi_org_10_5281_zenodo_20109260
institution Zenodo
language eng
publishDate 2026
publisher Zenodo
record_format zenodo
spellingShingle Cellular Inference Mesh: Salabim DES of edge-cloud LLM inference under PACELC saturation
Flores, Carlos Ulisses
cellular inference mesh
saturated pipeline conjecture
PACELC
edge-cloud LLM serving
speculative decoding
multilingual AGV
LLMOps
Mobile Edge Intelligence
salabim
discrete-event simulation
FAIR research software
reproducibility
Discrete-event simulation in Salabim of the Cellular Inference Mesh edge-cloud architecture for industrial multilingual AGV voice command processing under PACELC. Implements 300 Monte-Carlo replicates over 3 scenarios and 2 arms for 1800 simulated seconds, with parametric variation per replicate over realistic operational ranges. Verifies the Pareto-Efficiency Theorem (over 5 axes against 256 sub-compositions and 7 external architectures) and articulates the Saturated Pipeline Conjecture, which explains the structural insensitivity of percentile-99 latency to fallback-rate reductions under edge saturation. Reproducibility guaranteed via SHA-256 hash chain, seed_canonical=42, 53 pytest tests with bit-parity.
title Cellular Inference Mesh: Salabim DES of edge-cloud LLM inference under PACELC saturation
topic cellular inference mesh
saturated pipeline conjecture
PACELC
edge-cloud LLM serving
speculative decoding
multilingual AGV
LLMOps
Mobile Edge Intelligence
salabim
discrete-event simulation
FAIR research software
reproducibility
url https://doi.org/10.5281/zenodo.20109260