Multi-Armed Bandits for Spectrum Allocation in Multi-Agent Channel Bonding WLANs

Sergio Barrachina-Munoz,Alessandro Chiumento,Boris Bellalta

doi:10.1109/access.2021.3114430

Sergio Barrachina-Munoz, Alessandro Chiumento + Show 1 more

Open Access

https://doi.org/10.1109/access.2021.3114430

Copy DOI

Abstract

While dynamic channel bonding (DCB) is proven to boost the capacity of wireless local area networks (WLANs) by adapting the bandwidth on a per-frame basis, its performance is tied to the primary and secondary channel selection. Unfortunately, in uncoordinated high-density deployments where multiple basic service sets (BSSs) may potentially overlap, hand-crafted spectrum management techniques perform poorly given the complex hidden/exposed nodes interactions. To cope with such challenging Wi-Fi environments, in this paper, we first identify machine learning (ML) approaches applicable to the problem at hand and justify why model-free RL suits it the most. We then design a complete RL framework and call into question whether the use of complex RL algorithms helps the quest for rapid learning in realistic scenarios. Through extensive simulations, we derive that stateless RL in the form of lightweight multi-armed-bandits (MABs) is an efficient solution for rapid adaptation avoiding the definition of broad and/or meaningless states. In contrast to most current trends, we envision lightweight MABs as an appropriate alternative to the cumbersome and slowly convergent methods such as Q-learning, and especially, deep reinforcement learning.

Highlights

State-of-the-art wireless applications like augmented reality, virtual reality, or 8K video streaming are urging nextgeneration wireless local area networks (WLANs) to support ever-increasing demands on performance
This paper focuses on multi-armed bandits (MABs), a stateless RL formulation that enables faster online adaptation than temporal difference algorithms like Qlearning, and especially deep reinforcement learning (DRL)
We call into question whether the use of complex RL algorithms helps the quest and derive that stateless algorithms, especially in the form of lightweight MABs, are an efficient solution for rapid adaptation avoiding extensive or meaningless states

Summary

Introduction

State-of-the-art wireless applications like augmented reality, virtual reality, or 8K video streaming are urging nextgeneration wireless local area networks (WLANs) to support ever-increasing demands on performance. Among the different approaches to enhance spectrum efficiency in WLANs, channel bonding was introduced in 802.11n-2009 [2] for bonding up to 40 MHz and further extended in 802.11ac/ax [3], [4] and 802.11be [5] to bond up to 160 and 320 MHz, respectively, in the 5-GHz band. Up to this date, the standards allow static and dynamic channel bonding (DCB).

Objectives

Results

Conclusion