稀疏贝叶斯学习水声信道估计与脉冲噪声抑制方法  

Underwater acoustic channel estimation and impulsive noise mitigation based on sparse Bayesian learning

作　　者：殷敬伟 高新博[1,2,3] 韩笑 张晓 王大宇[5] 张锦灿 YIN Jingwei;GAO Xinbo;HAN Xiao;ZHANG Xiao;WANG Dayu;ZHANG Jincan(Acoustic Science and Technology Laboratory,Harbin Engineering University,Harbin 150001;Key Laboratory of Marine Information Acquisition and Security(Harbin Engineering University),Ministry of Industry and Information Technology,Harbin 150001;College of Underwater Acoustic Engineering,Harbin Engineering University,Harbin 150001;College of Computer Science and Technology,Jilin University,Changchun 130012;The 54th Research Institute of CETC,Shijiazhuang 050081)

机构地区：[1]哈尔滨工程大学水声技术重点实验室,哈尔滨150001 [2]工业和信息化部海洋信息获取与安全工信部重点实验室(哈尔滨工程大学),哈尔滨150001 [3]哈尔滨工程大学水声工程学院,哈尔滨150001 [4]吉林大学计算机科学与技术学院,长春130012 [5]中国电子科技集团公司第五十四研究所,石家庄050081

出　　处：《声学学报》2021年第6期813-824,共12页Acta Acustica

基　　金：国家自然科学基金项目(61901136,61631008);国家重点研发计划项目(2018YFC1405900);声呐技术重点实验室开放基金项目(6142109KF201802)资助。

摘　　要：水声信道具有显著的稀疏特性,利用稀疏贝叶斯学习(SBL)算法能够实现稀疏水声信道的有效估计。针对SBL计算复杂度较高的问题,将广义近似消息传递-稀疏贝叶斯学习(GAMP-SBL)引入水声信道估计。该方法在SBL的框架下结合GAMP以消息传递的方式计算信道冲激响应,能够有效降低SBL的计算复杂度。针对假设背景噪声服从高斯分布的信道估计方法在脉冲噪声环境下性能下降问题,提出了基于GAMP-SBL的脉冲噪声抑制水声信道估计方法:首先利用脉冲噪声时域稀疏特性,采用GAMP-SBL估计脉冲噪声并进行抑制,然后再次利用GAMP-SBL实现水声信道估计.基于第九次北极科考冰下脉冲噪声的两次仿真结果表明,所提出的方法在归一化均方误差上相对于未进行脉冲噪声抑制的GAMP-SBL最大分别降低了18.71%,6.61%,在信道解码前误码率上最大分别降低了1.66%,4.05%,并且相对于Clipping方法更加稳健。在信噪比为20 dB时,误码率可低于10^(-2)。It is well known that UnderWater Acoustic(UWA)channel is sparse.Sparse Bayesian Learning(SBL)can estimate sparse UWA Channel Impulsive Response(CIR)effectively.Considering the relatively high complexity of SBL,Generalized Approximate Message Passing-Sparse Bayesian Learning(GAMP-SBL)algorithm is incorporated into UWA channel estimation which estimates CIR with message passing in SBL framework and reduce the computational complexity of SBL without losing much performance.Under the environment with impulsive noise,the performance of channel estimation algorithms with the assumption of Gaussian distributed background noise will decrease.By exploiting the sparsity of impulsive noise in the time domain,a GAMP-SBL based channel estimation-impulsive noise mitigation method is proposed to improve the performance of channel estimation under the impulsive noise environment,in which GAMP-SBL is utilized to mitigate the impulsive noise and estimate UWA channel respectively.Simulation results from the 9th Chinese National Arctic Research Expedition verify that the proposed algorithm reduces Normalized Mean Square Error(NMSE)by 18.71%and 6.61%mostly,reduces Bit Error Rate(BER)by 1.66%and 4.05%mostly compared with GAMP-SBL.Besides,the proposed method is more robust than Clipping and BER is less than 10^(-2) in 20 dB Signal to Noise Ratio(SNR).

关 键 词：稀疏贝叶斯学习 水声信道估计 脉冲噪声抑制 归一化均方误差 稀疏水声信道 稀疏特性 信道解码 高斯分布 

分 类 号：TN929.3[电子电信—通信与信息系统] TP18[电子电信—信息与通信工程]