End-Edge Model Collaboration: Bandwidth Allocation for Data Upload and Model Transmission

2025-04-19

AI Summary

To address the inference accuracy degradation in bandwidth-constrained IoT-edge collaborative deployment of large models, this paper proposes a communication-computation co-designed joint bandwidth optimization framework. We first quantitatively model the coupled impact of data upload and model feedback on edge-side inference accuracy, formulating a convex accuracy-communication trade-off optimization problem. By integrating bandwidth-aware co-modeling with lightweight model distillation, our approach maximizes inference accuracy while satisfying strict latency constraints. Experiments across diverse bandwidth conditions and data scales demonstrate that the proposed method significantly improves edge-side mean Average Precision (mAP), achieving up to a 23.6% gain over state-of-the-art baselines. This work establishes a verifiable, bandwidth-aware allocation paradigm for deploying large models on resource-constrained edge-IoT systems.

Technology Category

• Machine Learning: Learning on the Edge & Model Compression
• Computer Vision: Large Vision Models
• Constraint Satisfaction and Optimization: Distributed CSP/Optimization

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Cloud, edge and content delivery systems for the Web
• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs

Abstract

The widespread adoption of large artificial intelligence (AI) models has enabled numerous applications of the Internet of Things (IoT). However, large AI models require substantial computational and memory resources, which exceed the capabilities of resource-constrained IoT devices. End-edge collaboration paradigm is developed to address this issue, where a small model on the end device performs inference tasks, while a large model on the edge server assists with model updates. To improve the accuracy of the inference tasks, the data generated on the end devices will be periodically uploaded to edge serve to update model, and a distilled model of the updated one will be transmitted back to the end device. Subjected to the limited bandwidth for the communication link between the end device and the edge server, it is important to investigate the system should allocate more bandwidth for data upload or model transmission. In this paper, we characterize the impact of data upload and model transmission on inference accuracy. Subsequently, we formulate a bandwidth allocation problem. By solving this problem, we derive an efficient optimization framework for the end-edge collaboration system. The simulation results demonstrate our framework significantly enhances mean average precision (mAP) under various bandwidths and datasizes.

Problem

• Optimize bandwidth allocation between data upload and model transmission
• Improve inference accuracy in end-edge collaboration systems
• Balance resource constraints for IoT devices and edge servers

Innovation

• End-edge collaboration for AI model efficiency
• Dynamic bandwidth allocation optimization framework
• Balanced data upload and model transmission