講演抄録/キーワード |
講演名 |
2019-02-21 15:25
User Association in mmWave Networks using Reinforcement Learning ○Yuva Kumar S.・Fereidoun H. Panahi・Tomoaki Ohtsuki(KeioUniv) |
抄録 |
(和) |
With the advent of the fifth generation (5G) systems, there is an increasing demand for high data rate transmission and serving the increased proliferation of mobile and connected devices. This, in turn, has seen spectrum crunch to serve the demand. One major solution is to move towards millimeter-wave (mmWave) frequencies due to its large spectral bandwidth. However, mmWave frequencies experience large propagation path loss and are very sensitive to blockages like human bodies and buildings. This gives rise to unstable connectivity and unreliable communication. In this report, we propose a User Association based on Reinforcement Learning (UARL) for a typical user equipment (UE) in mmWave networks to reduce the number of blockages experienced by the UE. We use reinforcement learning (RL), where the base station (BS) would be updated with the available and possible blockages for a UE. We consider an mmWave network where BSs are distributed in the landscape such that it satisfies path diversity for the UE at any given point, i.e., each UE will be served by secondary beam once the UE experiences blockage in the initial beam. Investigation of the performance of the UARL shows better performance in terms of the number of blockages experienced by the UEs than the mmWave network without RL. |
(英) |
With the advent of the fifth generation (5G) systems, there is an increasing demand for high data rate transmission and serving the increased proliferation of mobile and connected devices. This, in turn, has seen spectrum crunch to serve the demand. One major solution is to move towards millimeter-wave (mmWave) frequencies due to its large spectral bandwidth. However, mmWave frequencies experience large propagation path loss and are very sensitive to blockages like human bodies and buildings. This gives rise to unstable connectivity and unreliable communication. In this report, we propose a User Association based on Reinforcement Learning (UARL) for a typical user equipment (UE) in mmWave networks to reduce the number of blockages experienced by the UE. We use reinforcement learning (RL), where the base station (BS) would be updated with the available and possible blockages for a UE. We consider an mmWave network where BSs are distributed in the landscape such that it satisfies path diversity for the UE at any given point, i.e., each UE will be served by secondary beam once the UE experiences blockage in the initial beam. Investigation of the performance of the UARL shows better performance in terms of the number of blockages experienced by the UEs than the mmWave network without RL. |
キーワード |
(和) |
Millimeter-wave communication / blockage probability / reinforcement learning / / / / / |
(英) |
Millimeter-wave communication / blockage probability / reinforcement learning / / / / / |
文献情報 |
映情学技報, vol. 43, no. 6, BCT2019-29, pp. 25-28, 2019年2月. |
資料番号 |
BCT2019-29 |
発行日 |
2019-02-14 (BCT) |
ISSN |
Print edition: ISSN 1342-6893 Online edition: ISSN 2424-1970 |
PDFダウンロード |
|
|