摘要
针对传统空间调制系统结构优化及复杂度较高的缺陷,提出了一种基于自动编码器的空间调制系统设计方案(Autoencoder Based Spatial Modulation Multiple-Input Multiple-Output,AE-SM-MIMO)。该方案通过使用卷积神经网络来构建系统的编码器和解码器,进行空间调制的端到端学习,进而实现不同发射天线、接收天线以及调制方式下信息比特流的传输与接收,在系统误码性能和复杂度之间提供了较好的平衡。在瑞利衰落信道环境下,将所提出的基于卷积神经网络的AE-SM-MIMO系统方案与传统的空间调制系统进行性能对比,实验结果表明,AE-SM-MIMO系统方案可获得接近最大似然检测算法的误码性能,并表现出了良好的适应性和泛化能力。此外,相比于传统最大似然检测算法和最大比合并检测算法,在4发2收BPSK调制方式下AE-SM-MIMO系统方案需要更少的计算时间,分别减少了78%和26%,大大降低了系统的时间复杂度。
For the optimization of traditional spatial modulation system structure and the defects of high complexity,a spatial modulation system design scheme based on automatic encoder(AE-SM-MIMO)is proposed.In this scheme,the encoder and decoder of the system are constructed by using convolutional neural network,and the end-to-end learning of spatial modulation is carried out,so as to realize the transmission and reception of information bit streams under different transmitting antennas,receiving antennas and different modulation modes.The system provides a good balance between error performance and complexity.In Rayleigh fading channel environment,the performance of the proposed AE-SM-MIMO system scheme based on the convolutional neural network(CNN)is compared with that of the traditional spatial modulation system.The experimental results show that the AE-SM-MIMO system scheme can achieve the error performance close to the maximum likelihood detection algorithm,and shows good adaptability and generalization ability.In addition,compared with traditional maximum likelihood detection algorithm and maximum ratio combining detection algorithm,AE-SM-MIMO system scheme in the BPSK modulation mode with four transmitting antennas and two receiving antennas requires less computing time,which is reduced by 78%and 26%respectively,and greatly reduces the time complexity of the system.
作者
任嘉欣
王旭东
吴楠
REN Jiaxin;WANG Xudong;WU Nan(School of Computer and Software,Dalian Neusoft University of Information,Dalian 116023,China;College of Information and Technology,Dalian Maritime University,Dalian 116026,China)
出处
《电讯技术》
北大核心
2024年第11期1772-1779,共8页
Telecommunication Engineering
基金
国家自然科学基金资助项目(61801074)。
关键词
空间调制
自编码器
卷积神经网络
瑞利衰落信道
spatial modulation
autoencoder
convolutional neural network
Rayleigh fading channel
作者简介
任嘉欣,女,1998年生于辽宁大连,硕士研究生,主要研究方向为深度学习在通信系统中的应用;通信作者:王旭东,男,1967年生于黑龙江哈尔滨,2008年于西安电子科技大学获工学博士学位,现为教授,主要研究方向为光无线通信调制技术、可见光无线通信及定位技术、基于人工智能的无线通信系统设计,Email:wxd@dlmu.edu.cn;吴楠,男,1979年生于辽宁大连,2008年于英国南安普顿大学获工学博士学位,现为副教授,主要研究方向为现代移动无线通信系统、基于深度学习的无线通信系统、可见光通信系统(MIMO、OFDM、信道编码、协作通信、自组网络)等。
