Adaptive DNN Partitioning and Offloading in Heterogeneous Edge-Cloud Continuum

2026-05-10

AI Summary

This study addresses the limitations of existing static DNN partitioning and offloading approaches, which struggle to adapt to runtime dynamics in the edge–cloud continuum and lack validation on real hardware. To overcome these challenges, this work proposes a dynamic neural network layer partitioning framework that profiles model performance at launch, continuously monitors network conditions, and periodically re-evaluates the partitioning strategy to adapt to environmental changes. The approach is the first to demonstrate runtime-adaptive partitioning and offloading on a heterogeneous edge–fog–cloud testbed composed of Raspberry Pi devices, laptops, and high-performance PCs. Experimental results show that, compared to static baselines, the proposed method reduces energy consumption by 27.09%–35.82% and decreases end-to-end latency by 6.34%–22.92% across VGG16, AlexNet, and MobileNetV2 models.

Abstract

In recent years, the use of artificial intelligence on resource-constrained IoT devices has grown significantly. However, existing approaches to DNN partitioning and offloading across the edge-cloud continuum typically rely on static methods that ignore runtime dynamics. Furthermore, they are often evaluated in simulated environments rather than on real hardware. To address this gap, we propose a framework that dynamically splits neural network layers across the heterogeneous continuum. The framework profiles the model at startup, measures network link conditions between nodes, and periodically re-evaluates the partition to adapt to environmental changes. We created a physical testbed comprising a Raspberry Pi edge device, a laptop fog, and a high-performance desktop PC as the cloud. We evaluated the framework over three widely adopted convolutional neural networks: VGG16, AlexNet, and MobileNetV2. Our results show that the framework achieves reductions in energy and end-to-end latency of 27.09--35.82% and 6.34--22.92%, respectively, compared to a static partitioning baseline. These findings confirm the superiority of adaptive to static partitioning.

Problem

• DNN partitioning
• edge-cloud continuum
• runtime dynamics
• static offloading
• heterogeneous systems

Innovation

• adaptive DNN partitioning
• edge-cloud continuum
• dynamic offloading
• heterogeneous computing
• real-world testbed