Multi-dimensional Autoscaling of Processing Services: A Comparison of Agent-based Methods

2025-06-12

AI Summary

To address the failure of conventional autoscaling methods under stringent resource constraints in edge computing, this paper proposes a multidimensional adaptive scaling framework that jointly optimizes hardware resource allocation and internal service configurations (e.g., model inference parameters, input image resolution), significantly improving SLO compliance. We introduce a novel elasticity control paradigm integrating both software and hardware configuration dimensions. For the first time, we systematically evaluate four intelligent agents—Active Inference, Deep Q-Network (DQN), Analysis of Structural Knowledge (ASK), and Deep Active Inference—on real-world edge vision services: YOLOv8-based object detection and OpenCV-based QR code recognition. Experimental results show that ASK achieves the fastest convergence; DQN heavily relies on pretraining; Deep Active Inference balances theoretical rigor with engineering scalability; and all four agents attain acceptable SLO performance under dynamic edge workloads.

Technology Category

• Machine Learning: Learning on the Edge & Model Compression
• Multiagent Systems: Adversarial Agents
• Search and Optimization: Learning to Search

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Cloud, edge and content delivery systems for the Web
• Responsible Web: Machine-in-the-loop, human agency and autonomy
• Economics, Online Markets and Human Computation: Incentives in network design for Web infrastructures and ecosystems

Abstract

Edge computing breaks with traditional autoscaling due to strict resource constraints, thus, motivating more flexible scaling behaviors using multiple elasticity dimensions. This work introduces an agent-based autoscaling framework that dynamically adjusts both hardware resources and internal service configurations to maximize requirements fulfillment in constrained environments. We compare four types of scaling agents: Active Inference, Deep Q Network, Analysis of Structural Knowledge, and Deep Active Inference, using two real-world processing services running in parallel: YOLOv8 for visual recognition and OpenCV for QR code detection. Results show all agents achieve acceptable SLO performance with varying convergence patterns. While the Deep Q Network benefits from pre-training, the structural analysis converges quickly, and the deep active inference agent combines theoretical foundations with practical scalability advantages. Our findings provide evidence for the viability of multi-dimensional agent-based autoscaling for edge environments and encourage future work in this research direction.

Problem

• Dynamic adjustment of hardware and service configurations in edge computing
• Comparison of agent-based autoscaling methods for constrained environments
• Evaluation of multi-dimensional scaling agents on real-world processing services

Innovation

• Agent-based framework for multi-dimensional autoscaling
• Dynamic adjustment of hardware and service configurations
• Comparison of four scaling agents in edge environments