基于神经网络的低轨卫星多波束系统峰均比自适应限幅法  

An adaptive limiting method in LEO satellite multi-beam system based on neural network

作　　者：刘子健 姜泉江 刘会杰 LIU Zijian;JIANG Quanjiang;LIU Huijie(Innovation Academy for Microsatellites,Chinese Academy of Sciences,Shanghai 200120,China;University of Chinese Academy of Sciences,Beijing 100049,China;ShanghaiTech University,Shanghai 201210,China)

机构地区：[1]中国科学院微小卫星创新研究院,上海201203 [2]中国科学院大学,北京100049 [3]上海科技大学,上海201210

出　　处：《中国科学院大学学报（中英文）》2023年第5期670-676,共7页Journal of University of Chinese Academy of Sciences

基　　金：国家自然科学基金重大研究计划重点支持项目(91738201);上海市科技创新行动计划(20511106602)资助。

摘　　要：针对低轨卫星多波束系统在多方向通信时出现峰均比过高的问题,提出一种基于深度神经网络的峰均比抑制方法。该方法主要根据多波束通信系统在不同场景下的目标方向位置、接收端信噪比、误码率误差范围等输入层参数,自适应地选择限幅法的最优门限值,从而对每个阵元上合成的信号进行限幅操作,在保证低轨卫星多波束系统误码率的前提下降低了该系统的峰均比。最后通过仿真验证,该方法比传统固定门限的限幅法在误码率误差范围内对峰均比有明显改善。Aiming at the problem that the peak-to-average ratio is too high when the LEO satellite multi-beam system communicates with multiple target directional angles,a method for suppressing the peak-to-average ratio based on a deep neural network is proposed.This method can adaptively select the optimal threshold of the limiting method based on the input layer parameters such as the target direction angle position of the multi-beam communication system,the SNR of the receiving end,and the error range of the BER.The signal synthesized on the element is subjected to amplitude limiting operation,which reduces the peak-to-average ratio of the LEO satellite multibeam system while ensuring the BER of the system.Finally,it is verified by simulation that this method can significantly improve the peak-to-average ratio within the error range of the BER compared with the traditional fixed-threshold limiting method.

关 键 词：低轨卫星多波束 峰均比 神经网络 限幅法 

分 类 号：TN927[电子电信—通信与信息系统]