QoS-Aware Load Balancing in the Computing Continuum via Multi-Player Bandits

2025-12-21

AI Summary

To address the challenge of client-granular QoS assurance across the computing continuum, this paper proposes QEdgeProxy—a lightweight, decentralized load balancer acting as an intermediary between IoT devices and service instances. To tackle the frequent neglect of per-client latency and reliability requirements in dynamic environments, we introduce the first integration of heterogeneous-reward multi-player multi-armed bandits (MP-MAB) with kernel density estimation (KDE) for nonparametric modeling of QoS success probability; we further design an adaptive exploration mechanism to rapidly respond to non-stationary traffic patterns and abrupt performance shifts. Evaluation on a K3s-based edge testbed under realistic network conditions and edge AI workloads demonstrates that QEdgeProxy significantly improves per-client QoS satisfaction rates compared to proximity-based scheduling and reinforcement learning baselines, while efficiently handling traffic surges and dynamic service instance scaling.

Technology Category

• Machine Learning: Learning on the Edge & Model Compression
• Multiagent Systems: Mechanism Design
• Planning, Routing, and Scheduling: Scheduling under Uncertainty

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Energy management for devices in mobile Web and WoT environments
• Economics, Online Markets and Human Computation: Incentives in network design for Web infrastructures and ecosystems
• Graph Algorithms and Modeling for the Web: Efficient manipulation of static and dynamic Web-related graphs

Abstract

As computation shifts from the cloud to the edge to reduce processing latency and network traffic, the resulting Computing Continuum (CC) creates a dynamic environment where it is challenging to meet strict Quality of Service (QoS) requirements and avoid service instance overload. Existing methods often prioritize global metrics, overlooking per-client QoS, which is crucial for latency-sensitive and reliability-critical applications. We propose QEdgeProxy, a decentralized QoS-aware load balancer that acts as a proxy between IoT devices and service instances in CC. We formulate the load balancing problem as a Multi-Player Multi-Armed Bandit (MP-MAB) with heterogeneous rewards, where each load balancer autonomously selects service instances that maximize the probability of meeting its clients' QoS targets by using Kernel Density Estimation (KDE) to estimate QoS success probabilities. It also incorporates an adaptive exploration mechanism to recover rapidly from performance shifts and non-stationary conditions. We present a Kubernetes-native QEdgeProxy implementation and evaluate it on an emulated CC testbed deployed on a K3s cluster with realistic network conditions and a latency-sensitive edge-AI workload. Results show that QEdgeProxy significantly outperforms proximity-based and reinforcement-learning baselines in per-client QoS satisfaction, while adapting effectively to load surges and instance availability changes.

Problem

• Decentralized load balancing for IoT devices in Computing Continuum
• Meeting per-client QoS targets in dynamic, latency-sensitive environments
• Adapting to load surges and instance changes via multi-player bandits

Innovation

• Decentralized QoS-aware load balancer using MP-MAB
• Kernel Density Estimation for QoS success probability
• Adaptive exploration for dynamic performance recovery