Proactive Dynamic Channel Selection Based on Multi-Armed Bandit Learning for 5G NR-U
With an increasing demand of mobile data traffic in fifth-generation (5G) wireless communication systems, new radio-unlicensed (NR-U) technology has been regarded as a promising technology to address the exponential growth of data traffic by offloading the traffic to unlicensed bands. Nevertheless,...
Gespeichert in:
Veröffentlicht in: | IEEE access 2020, Vol.8, p.196363-196374 |
---|---|
Hauptverfasser: | , , , |
Format: | Artikel |
Sprache: | eng |
Schlagworte: | |
Online-Zugang: | Volltext |
Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
Zusammenfassung: | With an increasing demand of mobile data traffic in fifth-generation (5G) wireless communication systems, new radio-unlicensed (NR-U) technology has been regarded as a promising technology to address the exponential growth of data traffic by offloading the traffic to unlicensed bands. Nevertheless, how to efficiently share the unlicensed spectrum resource among the NR and Wi-Fi systems is a key challenge to be addressed, especially in a dynamic network environment. In this article, we investigate a distributed channel access mechanism and focus on the channel selection for NR-U users to decide the optimal unlicensed channel for uplink traffic offloading. We formulate the selection problem as a non-cooperative game, which is proven to be an exact potential game. However, the Nash equilibrium (NE) point is hard to achieve, due to the unknown dynamic environment. Based on multi-armed bandit learning techniques, an online learning distributed channel selection algorithm (OLDCSA) is proposed and proven to have similar performance to the NE point. Finally, simulation results reveal that our proposed algorithm outperforms the existing random selection by 16.45% on average and is close to the exhaustive search in the dynamic unknown environment. |
---|---|
ISSN: | 2169-3536 2169-3536 |
DOI: | 10.1109/ACCESS.2020.3034360 |