BouquetFL: Emulating diverse participant hardware in Federated Learning

Arno Geimer · 2026

Read BouquetFL: Emulating diverse participant hardware in Federated Learning by Arno Geimer (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2602-06498.txt.

Description

Federated learning splits training across user devices, but devices differ wildly in speed and memory. Arno Geimer proposes a realistic test framework that emulates that spread on one machine. Small real-device trials are hard to scale and repeat, while trace or probability simulations carry compounding uncertainties.

The framework reproduces each client's compute capacity and memory limits at run time, capturing how those limits interact. Unlike fixed traces, it is workload-aware: the same device group can run and fail differently depending on the task. Across several workloads, it preserved relative device performance and matched hardware behavior, giving researchers a repeatable way to judge hardware-aware methods.

Catalog record

CREATORArno Geimer
YEAR2026
COLLECTIONPAPERS
FORMATTXT
FILE SIZE13 KB
VERSION1.0
LICENSEhttp://creativecommons.org/licenses/by/4.0/
RIGHTSCC BY
SOURCEarXiv

Source: https://arxiv.org/abs/2602-06498

Browse the full library · About the Tower of Babel