.png){kind=icon}
Return
End-to-End Learning for RIS-Aided Communication Systems
.png){kind=icon}
PRE
AI
DOI：10.1109/TVT.2022.3162624.png)
.png){kind=icon}
Original request for help.png){kind=icon}
Share.png){kind=icon}
Save
Abstract
En 中文
From adapting to the propagation environment to reconstructing the propagation environment, reconfigurable intelligent surface (RIS) changes the design paradigm of wireless communications. To reconstruct the propagation environment, joint beamforming of RIS and multi-input multi-output (MIMO) is crucial. Unfortunately, due to the coupling effect of the active beamforming of MIMO and passive beamforming of RIS, it is difficult to find the optimal solution to the joint beamforming problem, so a serious performance loss will be caused. In this paper, inspired by the end-to-end (E2E) learning of communication system, we propose the E2E learning based RIS-aided communication system to mitigate the performance loss via deep learning techniques. The key idea is to simultaneously optimize the signal processing functions at base station (BS), RIS, and user, including active beamforming for BS and user, passive beamforming for RIS. This way is able to avoid the performance loss caused by alternately optimizing each function of the RIS-aided system. Specifically, we firstly utilize a deep neural network (DNN) to realize the modulation and beamforming for BS and utilize another DNN to realize the demodulation and combining for user. Then, the RIS passive beamforming is also represented by trainable parameters, which could be simultaneously optimized with the DNNs at BS and user. Simulation results show that the proposed E2E learning based RIS-aided communication system could achieve the better bit error rate (BER) performance than traditional RIS-aided communication systems.
Keywords:
Array signal processing
Communication systems
Transmitters
Receivers
Optimized production technology
Bit error rate
MIMO communication
End-to-end learning
reconfigurable intelligent surface
bit error rate
deep neural network
Journal
IF:
7.1
Papers:
1.8W
Citations:
6.6W
Organization
Cited Papers
Neuropsychiatric Features of Frontotemporal Dementia: Evaluation of Consensus Criteria and Review
.png){kind=icon}
PREAI
ShareSaveDesempenho de crianças do ensino fundamental na solução de problemas aritméticos<A NAME="top"></A>
OAAI
OAAIShareSave
ShareSaveShareSaveShareSaveShareSaveShareSave
View more