Going Beyond the Edge: Distributed Inference of Transformer Models on Ultra-Low-Power Wireless Devices

Fuente: arXiv
Saved in:
Bibliographic Details
Main Authors: Gräfe, Alexander, Huo, Ding, de Bakker, Vincent, Berger, Johannes, Zimmerling, Marco, Trimpe, Sebastian
Format: Preprint
Published: 2026
Subjects:
Online Access:
Tags: Add Tag
No Tags, Be the first to tag this record!
_version_ 1866911693702955008
author Gräfe, Alexander
Huo, Ding
de Bakker, Vincent
Berger, Johannes
Zimmerling, Marco
Trimpe, Sebastian
author_facet Gräfe, Alexander
Huo, Ding
de Bakker, Vincent
Berger, Johannes
Zimmerling, Marco
Trimpe, Sebastian
contents Transformer models are rapidly becoming a cornerstone of modern Internet of Things (IoT) applications, yet their computational and memory demands far exceed the capabilities of a single typical ultra-low-power IoT device. We present CATS, a framework for distributed transformer inference on ultra-low-power wireless devices, enabling multiple devices to collaboratively execute models far larger than what a single device can sustain. At its core, CATS is a communication-aware distributed transformer inference scheme co-designed across transformer partitioning, wireless communication and training. It employs SomeGather, a new pruned communication primitive that selectively broadcasts activation columns to reduce communication bandwidth and RAM usage without sacrificing model accuracy. Building on SomeGather, we design a partitioning method that exploits this primitive for efficient model parallelism. To cope with unreliable wireless communication, CATS employs message-dropout during training, which mimics packet losses and yields models that are robust to message loss during inference. In real-world experiments, we show that CATS brings distributed transformer inference to ultra-low-power wireless devices for the first time, with deployments on up to 16 devices that collaboratively execute transformer models up to 14 times larger than what a single device can run.
format Preprint
id arxiv_https___arxiv_org_abs_2605_15694
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle Going Beyond the Edge: Distributed Inference of Transformer Models on Ultra-Low-Power Wireless Devices
Gräfe, Alexander
Huo, Ding
de Bakker, Vincent
Berger, Johannes
Zimmerling, Marco
Trimpe, Sebastian
Machine Learning
Transformer models are rapidly becoming a cornerstone of modern Internet of Things (IoT) applications, yet their computational and memory demands far exceed the capabilities of a single typical ultra-low-power IoT device. We present CATS, a framework for distributed transformer inference on ultra-low-power wireless devices, enabling multiple devices to collaboratively execute models far larger than what a single device can sustain. At its core, CATS is a communication-aware distributed transformer inference scheme co-designed across transformer partitioning, wireless communication and training. It employs SomeGather, a new pruned communication primitive that selectively broadcasts activation columns to reduce communication bandwidth and RAM usage without sacrificing model accuracy. Building on SomeGather, we design a partitioning method that exploits this primitive for efficient model parallelism. To cope with unreliable wireless communication, CATS employs message-dropout during training, which mimics packet losses and yields models that are robust to message loss during inference. In real-world experiments, we show that CATS brings distributed transformer inference to ultra-low-power wireless devices for the first time, with deployments on up to 16 devices that collaboratively execute transformer models up to 14 times larger than what a single device can run.
title Going Beyond the Edge: Distributed Inference of Transformer Models on Ultra-Low-Power Wireless Devices
topic Machine Learning
url https://arxiv.org/abs/2605.15694