TY - JOUR
T1 - Long-Term CSI-based Design for RIS-Aided Multiuser MISO Systems Exploiting Deep Reinforcement Learning
AU - Ren, Hong
AU - Pan, Cunhua
AU - Wang, Liang
AU - Liu, Wang
AU - Kou, Zhoubin
AU - Wang, Kezhi
N1 - Funding information:
This work was supported in part by the National Key Research and Development Project under Grant 2019YFE0123600, National Natural Science Foundation of China (62101128) and Basic Research Project of Jiangsu Provincial Department of Science and Technology (BK20210205).
PY - 2022/3/10
Y1 - 2022/3/10
N2 - In this paper, we study the transmission design for reconfigurable intelligent surface (RIS)-aided multiuser communication networks. Different from most of the existing contributions, we consider long-term CSI-based transmission design, where both the beamforming vectors at the base station (BS) and the phase shifts at the RIS are designed based on long-term CSI, which can significantly reduce the channel estimation overhead. Due to the lack of explicit ergodic data rate expression, we propose a novel deep deterministic policy gradient (DDPG) based algorithm to solve the optimization problem, which was trained by using the channel vectors generated in an offline manner. Simulation results demonstrate that the achievable net throughput is higher than that achieved by the conventional instantaneous-CSI based scheme when taking the channel estimation overhead into account.
AB - In this paper, we study the transmission design for reconfigurable intelligent surface (RIS)-aided multiuser communication networks. Different from most of the existing contributions, we consider long-term CSI-based transmission design, where both the beamforming vectors at the base station (BS) and the phase shifts at the RIS are designed based on long-term CSI, which can significantly reduce the channel estimation overhead. Due to the lack of explicit ergodic data rate expression, we propose a novel deep deterministic policy gradient (DDPG) based algorithm to solve the optimization problem, which was trained by using the channel vectors generated in an offline manner. Simulation results demonstrate that the achievable net throughput is higher than that achieved by the conventional instantaneous-CSI based scheme when taking the channel estimation overhead into account.
KW - Array signal processing
KW - Channel estimation
KW - Coherence time
KW - deep reinforcement learning
KW - intelligent reflecting surface (IRS)
KW - Interference
KW - Precoding
KW - Reconfigurable intelligent surface (RIS)
KW - Rician channels
KW - Signal to noise ratio
UR - http://www.scopus.com/inward/record.url?scp=85122586218&partnerID=8YFLogxK
U2 - 10.1109/LCOMM.2021.3140155
DO - 10.1109/LCOMM.2021.3140155
M3 - Article
AN - SCOPUS:85122586218
SN - 1089-7798
VL - 26
SP - 567
EP - 571
JO - IEEE Communications Letters
JF - IEEE Communications Letters
IS - 3
ER -