基于深度学习的水下无线光通信信噪比改善研究与实现  

Research and implementation of signal to noise ratio improvement of underwater free space optical communication based on deep learning

作　　者：孙猷 张俊杰[1] SUN You;ZHANG Junjie(Key Laboratory of Specialty Fiber Optics and Optical Access Networks,Shanghai University,Shanghai 200072,China)

机构地区：[1]上海大学特种光纤与光接入网重点实验室,上海200072

出　　处：《光通信技术》2021年第1期10-15,共6页Optical Communication Technology

基　　金：国家自然科学基金项目(批准号:61420106011、61601279、61601277)资助。

摘　　要：复杂的水下环境造成了水下光信号的质量下降,为提升水下无线光通信系统信号信噪比,结合深度神经网络提出了一种信噪比改善方法,该方法通过信号频谱有效抑制了信号噪声。实验结果表明:针对1 m传输距离的水下无线光通信16阶正交振幅调制-正交频分复用(16QAM-OFDM)信号,该方法可以实现约17 dB的信噪比提升同时误码率降低至1.709×10-3,可应用于水下无线光通信系统中提升传输性能,且具备一定的泛化能力。The complex underwater environment results in the degradation of underwater optical signal quality.In order to improve the signal to noise ratio of underwater free space optical communication system,a signal to noise ratio improvement method based on deep neural network is proposed,which effectively suppresses the signal noise through the signal spectrum.For the underwater free space optical communication 16 order quadrature amplitude modulation-orthogonal frequency division multiplexing(16QAM-OFDM)signal with 1 m transmission distance,the experimental results show that the proposed scheme can improve the signal to noise ratio of about 17 dB and reduce the bit error rate to 1.709×10-3,which can be applied to under-water free space optical communication system to improve the transmission performance and has certain generalization ability.

关 键 词：水下无线光通信 信噪比改善 深度学习 正交频分复用 

分 类 号：TN91[电子电信—通信与信息系统]