面向大规模MIMO CSI反馈的聚合网络
信息论
2021-05-05 v2 人工智能
信号处理
math.IT
摘要
在频分双工(FDD)模式下，有必要将信道状态信息(CSI)从用户设备发送至基站。下行CSI对于大规模多输入多输出(MIMO)系统获取潜在增益至关重要。近来，深度学习被广泛采用于大规模MIMO CSI反馈任务，且相比传统压缩感知方法被证明有效。在本文中，设计了一种名为ACRNet的新型网络，通过网络聚合与参数化RuLU激活提升反馈性能。此外，首次在CSI反馈任务中讨论了以换取更优性能来扩展网络架构的有效途径。实验表明，ACRNet在无需任何额外信息下优于大量先前最先进的反馈网络。
引用
@article{arxiv.2101.06618,
title = {Aggregated Network for Massive MIMO CSI Feedback},
author = {Zhilin Lu and Hongyi He and Zhengyang Duan and Jintao Wang and Jian Song},
journal= {arXiv preprint arXiv:2101.06618},
year = {2021}
}
备注
This version is only a draft of the final paper `Binarized Aggregated Network with Quantization: Flexible Deep Learning Deployment for CSI Feedback in Massive MIMO System`, which has been uploaded as arXiv:2105.00354. This incomplete version has some performance error, therefore it should be withdrawn