Optimizing Split Learning Latency in TinyML-Based IoT Systems
Fuente:
arXiv
Saved in:
| Main Authors: | , , , |
|---|---|
| Format: | Preprint |
| Published: |
2025
|
| Subjects: | |
| Online Access: | |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866910192990420992 |
|---|---|
| author | Jenhani, Zied Bensalem, Mounir Dizdarević, Jasenka Jukan, Admela |
| author_facet | Jenhani, Zied Bensalem, Mounir Dizdarević, Jasenka Jukan, Admela |
| contents | Split learning (SL) addresses the limitation of running deep learning inference directly on low-power edge/IoT nodes, in which it executes part of the inference process on the sensor and offloading the remainder to a companion device. Despite its promise, the inference latency of SL on constrained hardware under realistic low-power wireless protocols remains unexplored. This paper presents the first experimental latency benchmark of TinyML-based SL on ESP32-S3 boards, comparing four wireless communication protocol solutions (UDP, TCP, ESP-NOW, BLE). We also analyze the impact of the choice of different split points across different models (MobileNet-V2 and ResNet50) in terms of communication and computation overhead as a way to minimize the end-to-end inference latency. We propose a Beam Search-based algorithm for split point optimization that minimizes end-to-end latency, and compare it with other methods, including Greedy Search, First-Fit, Random-Fit, and Brute Force. ESP-NOW achieves the best RTT (3.6 s) and serves as the base protocol for the algorithm, which delivers near-optimal latency with processing time of 0.1 s for 5 devices. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2507_16594 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Optimizing Split Learning Latency in TinyML-Based IoT Systems Jenhani, Zied Bensalem, Mounir Dizdarević, Jasenka Jukan, Admela Networking and Internet Architecture Artificial Intelligence Distributed, Parallel, and Cluster Computing Split learning (SL) addresses the limitation of running deep learning inference directly on low-power edge/IoT nodes, in which it executes part of the inference process on the sensor and offloading the remainder to a companion device. Despite its promise, the inference latency of SL on constrained hardware under realistic low-power wireless protocols remains unexplored. This paper presents the first experimental latency benchmark of TinyML-based SL on ESP32-S3 boards, comparing four wireless communication protocol solutions (UDP, TCP, ESP-NOW, BLE). We also analyze the impact of the choice of different split points across different models (MobileNet-V2 and ResNet50) in terms of communication and computation overhead as a way to minimize the end-to-end inference latency. We propose a Beam Search-based algorithm for split point optimization that minimizes end-to-end latency, and compare it with other methods, including Greedy Search, First-Fit, Random-Fit, and Brute Force. ESP-NOW achieves the best RTT (3.6 s) and serves as the base protocol for the algorithm, which delivers near-optimal latency with processing time of 0.1 s for 5 devices. |
| title | Optimizing Split Learning Latency in TinyML-Based IoT Systems |
| topic | Networking and Internet Architecture Artificial Intelligence Distributed, Parallel, and Cluster Computing |
| url | https://arxiv.org/abs/2507.16594 |