Service Placement in Small Cell Networks Using Distributed Best Arm Identification in Linear Bandits

2025-06-22

AI Summary

To address high latency caused by cloud offloading for computation-intensive services in small-cell networks, this paper proposes a distributed multi-agent linear bandit framework for joint service deployment optimization across edge (small base stations, SBSs) and cloud. We design a novel distributed adaptive best-arm identification algorithm based on fixed confidence, enabling collaborative learning among multiple SBSs to rapidly converge to a near-optimal deployment policy under unknown, dynamic demand. Theoretically, the algorithm’s sample complexity decreases linearly with the number of SBSs, significantly improving communication efficiency. Simulations demonstrate substantial reductions in required learning rounds under target confidence levels and achieve high speedup ratios. Our key contribution is the first integration of linear structural priors with distributed best-arm identification—effectively balancing modeling accuracy, learning efficiency, and stringent edge resource constraints.

Technology Category

• Machine Learning: Online Learning & Bandits
• Search and Optimization: Distributed Search
• Multiagent Systems: Multiagent Learning

Application Category

• Systems and Infrastructure for Web, Mobile and WoT: Experiences and lessons learnt from Web-based algorithms and system deployments
• Economics, Online Markets and Human Computation: Incentives in network design for Web infrastructures and ecosystems
• Responsible Web: Human-perceived consequences of algorithmic deployment on the web

Abstract

As users in small cell networks increasingly rely on computation-intensive services, cloud-based access often results in high latency. Multi-access edge computing (MEC) mitigates this by bringing computational resources closer to end users, with small base stations (SBSs) serving as edge servers to enable low-latency service delivery. However, limited edge capacity makes it challenging to decide which services to deploy locally versus in the cloud, especially under unknown service demand and dynamic network conditions. To tackle this problem, we model service demand as a linear function of service attributes and formulate the service placement task as a linear bandit problem, where SBSs act as agents and services as arms. The goal is to identify the service that, when placed at the edge, offers the greatest reduction in total user delay compared to cloud deployment. We propose a distributed and adaptive multi-agent best-arm identification (BAI) algorithm under a fixed-confidence setting, where SBSs collaborate to accelerate learning. Simulations show that our algorithm identifies the optimal service with the desired confidence and achieves near-optimal speedup, as the number of learning rounds decreases proportionally with the number of SBSs. We also provide theoretical analysis of the algorithm's sample complexity and communication overhead.

Problem

• Optimizing service placement in small cell networks to reduce latency
• Balancing edge and cloud deployment under dynamic demand
• Distributed learning for efficient best-arm identification in bandits

Innovation

• Distributed multi-agent best-arm identification algorithm
• Linear bandit modeling for service demand
• Collaborative SBSs to reduce learning rounds