Adaptive DNN Partitioning and Offloading in Heterogeneous Edge-Cloud Continuum

TL;DR

This paper presents a dynamic framework for splitting neural networks across edge and cloud devices, adapting to runtime conditions to improve energy efficiency and latency.

Contribution

It introduces a real hardware testbed and a profiling-based adaptive partitioning method that outperforms static approaches in real-world scenarios.

Findings

Achieves 27.09--35.82% energy reduction

Reduces end-to-end latency by 6.34--22.92%

Demonstrates superiority over static partitioning in real hardware

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.