dorsal/arxiv
View SchemaWhere to Split? A Pareto-Front Analysis of DNN Partitioning for Edge Inference
| Authors | Adiba Masud, Nicholas Foley, Pragathi Durga Rajarajan, Palden Lama |
|---|---|
| Categories | |
| ArXiv ID | 2601.08025vv1 |
| URL | https://arxiv.org/abs/2601.08025 |
| License | http://arxiv.org/licenses/nonexclusive-distrib/1.0/ |
Abstract
The deployment of deep neural networks (DNNs) on resource-constrained edge devices is frequently hindered by their significant computational and memory requirements. While partitioning and distributing a DNN across multiple devices is a well-established strategy to mitigate this challenge, prior research has largely focused on single-objective optimization, such as minimizing latency or maximizing throughput. This paper challenges that view by reframing DNN partitioning as a multi-objective optimization problem. We argue that in real-world scenarios, a complex trade-off between latency and throughput exists, which is further complicated by network variability. To address this, we introduce ParetoPipe, an open-source framework that leverages Pareto front analysis to systematically identify optimal partitioning strategies that balance these competing objectives. Our contributions are threefold: we benchmark pipeline partitioned inference on a heterogeneous testbed of Raspberry Pis and a GPU-equipped edge server; we identify Pareto-optimal points to analyze the latency-throughput trade-off under varying network conditions; and we release a flexible, open-source framework to facilitate distributed inference and benchmarking. This toolchain features dual communication backends, PyTorch RPC and a custom lightweight implementation, to minimize overhead and support broad experimentation.
{
"annotation_id": "49f027a9-74d1-45fb-a76c-a3ccaad3c4dd",
"date_created": "2026-02-17T05:53:15.046000Z",
"date_modified": "2026-02-17T05:53:15.046000Z",
"file_hash": "b6ce2cc28abf93b566fada57054b430a77693642c8f8205c6069495d9dad451a",
"private": false,
"record": {
"abstract": "The deployment of deep neural networks (DNNs) on resource-constrained edge devices is frequently hindered by their significant computational and memory requirements. While partitioning and distributing a DNN across multiple devices is a well-established strategy to mitigate this challenge, prior research has largely focused on single-objective optimization, such as minimizing latency or maximizing throughput. This paper challenges that view by reframing DNN partitioning as a multi-objective optimization problem. We argue that in real-world scenarios, a complex trade-off between latency and throughput exists, which is further complicated by network variability. To address this, we introduce ParetoPipe, an open-source framework that leverages Pareto front analysis to systematically identify optimal partitioning strategies that balance these competing objectives.\n Our contributions are threefold: we benchmark pipeline partitioned inference on a heterogeneous testbed of Raspberry Pis and a GPU-equipped edge server; we identify Pareto-optimal points to analyze the latency-throughput trade-off under varying network conditions; and we release a flexible, open-source framework to facilitate distributed inference and benchmarking. This toolchain features dual communication backends, PyTorch RPC and a custom lightweight implementation, to minimize overhead and support broad experimentation.",
"arxiv_id": "2601.08025",
"authors": [
"Adiba Masud",
"Nicholas Foley",
"Pragathi Durga Rajarajan",
"Palden Lama"
],
"categories": [
"cs.DC"
],
"license": "http://arxiv.org/licenses/nonexclusive-distrib/1.0/",
"title": "Where to Split? A Pareto-Front Analysis of DNN Partitioning for Edge Inference",
"url": "https://arxiv.org/abs/2601.08025",
"version": "v1"
},
"schema_id": "dorsal/arxiv",
"source": {
"execution_id": "793439dd-39b7-4e83-9753-562f546db873",
"id": "arXiv Dataset",
"type": "Model",
"variant": "snapshot-2026-01-17",
"version": "0.1.0"
},
"user_id": 1000002
}