联合深度学习与FDSS的抑制PAPR研究  

Inhibition of PAPR by Combining Deep Learning and FDSS

作　　者：张金灿 杨鸿文[1] 

机构地区：[1]北京邮电大学信息与通信工程学院,北京

出　　处：《无线通信》2024年第4期51-59,共9页Hans Journal of Wireless Communications

摘　　要：OFDM是5G物理层关键技术之一,其缺点是PAPR过高,容易导致功放效率下降并造成信号失真。如何抑制OFDM信号的PAPR对低功耗的物联网终端来说是一个重要问题。本文提出了一种联合深度学习与FDSS的抑制PAPR算法。仿真结果表明,所提算法对于多种调制方式及子载波个数配置均有很好的PAPR抑制效果。在峰值功率受限的条件下,采用所提算法能使信道的传输增益提升6 dB左右。OFDM, one of the key techniques of the 5G physical layer, has the disadvantage of excessively high PAPR. The excessively high PAPR will lead to a decrease in power amplifier efficiency and cause signal distortion. How to suppress the PAPR of OFDM signals is an important problem for low-power Internet of Things terminals. This paper proposes a joint method combining deep learning and FDSS for PAPR suppression based on the PAPR suppression scheme of FDSS, and conducts simulation verification. The results show that the proposed joint method achieves excellent PAPR suppression performance in different modulation scenarios and different subcarrier numbers. Under the condition of peak power constraint, the proposed joint method can improve the transmission gain of the channel by about 6 dB.

关 键 词：峰均功率比 深度学习 频域赋形 

分 类 号：TN9[电子电信—信息与通信工程]