For bistatic multiple-input multiple-output(MIMO)radar,this paper presents a robust and direction finding method in strong impulse noise environment.By means of a new lower order covariance,the method is effective in ...For bistatic multiple-input multiple-output(MIMO)radar,this paper presents a robust and direction finding method in strong impulse noise environment.By means of a new lower order covariance,the method is effective in suppressing impulse noise and achieving superior direction finding performance using the maximum likelihood(ML)estimation method.A quantum equilibrium optimizer algorithm(QEOA)is devised to resolve the corresponding objective function for efficient and accurate direc-tion finding.The results of simulation reveal the capability of the presented method in success rate and root mean square error over existing direction-finding methods in different application situations,e.g.,locating coherent signal sources with very few snapshots in strong impulse noise.Other than that,the Cramér-Rao bound(CRB)under impulse noise environment has been drawn to test the capability of the presented method.展开更多
In order to solve the problem that the performance of traditional localization methods for mixed near-field sources(NFSs)and far-field sources(FFSs)degrades under impulsive noise,a robust and novel localization method...In order to solve the problem that the performance of traditional localization methods for mixed near-field sources(NFSs)and far-field sources(FFSs)degrades under impulsive noise,a robust and novel localization method is proposed.After eliminating the impacts of impulsive noise by the weighted out-lier filter,the direction of arrivals(DOAs)of FFSs can be estimated by multiple signal classification(MUSIC)spectral peaks search.Based on the DOAs information of FFSs,the separation of mixed sources can be performed.Finally,the estimation of localizing parameters of NFSs can avoid two-dimension spectral peaks search by decomposing steering vectors.The Cramer-Rao bounds(CRB)for the unbiased estimations of DOA and range under impulsive noise have been drawn.Simulation experiments verify that the proposed method has advantages in probability of successful estimation(PSE)and root mean square error(RMSE)compared with existing localization methods.It can be concluded that the proposed method is effective and reliable in the environment with low generalized signal to noise ratio(GSNR),few snapshots,and strong impulse.展开更多
为了解决冲击噪声下长短时记忆(long short term memory,LSTM)神经网络调制信号识别方法抗冲击噪声能力弱和超参数难以确定的问题,本文提出了一种演化长短时记忆神经网络的调制识别方法。利用基于短时傅里叶变换的卷积神经网络(convolut...为了解决冲击噪声下长短时记忆(long short term memory,LSTM)神经网络调制信号识别方法抗冲击噪声能力弱和超参数难以确定的问题,本文提出了一种演化长短时记忆神经网络的调制识别方法。利用基于短时傅里叶变换的卷积神经网络(convolution neural network,CNN)去噪模型对数据集去噪;结合量子计算机制和旗鱼优化器(sailfish optimizer,SFO)设计了量子旗鱼算法(quantum sailfish algorithm,QSFA)去演化LSTM神经网络以获得最优的超参数;使用演化长短时记忆神经网络作为分类器进行自动调制信号识别。仿真结果表明,采用所设计的CNN去噪和演化长短时记忆神经网络模型,识别准确率有了大幅度的提高。量子旗鱼算法演化LSTM神经网络模型降低了传统LSTM神经网络容易陷于局部极小值或者过拟合的概率,当混合信噪比为0 dB,所提方法对11种调制信号的平均识别准确率达到90%以上。展开更多
基金This work was supported by the National Natural Science Foundation of China(62073093)the Postdoctoral Scientific Research Developmental Fund of Heilongjiang Province(LBH-Q19098)+1 种基金the Heilongjiang Provincial Natural Science Foundation of China(LH2020F017)the Key Laboratory of Advanced Marine Communication and Information Technology,Ministry of Industry and Information Technology.
文摘For bistatic multiple-input multiple-output(MIMO)radar,this paper presents a robust and direction finding method in strong impulse noise environment.By means of a new lower order covariance,the method is effective in suppressing impulse noise and achieving superior direction finding performance using the maximum likelihood(ML)estimation method.A quantum equilibrium optimizer algorithm(QEOA)is devised to resolve the corresponding objective function for efficient and accurate direc-tion finding.The results of simulation reveal the capability of the presented method in success rate and root mean square error over existing direction-finding methods in different application situations,e.g.,locating coherent signal sources with very few snapshots in strong impulse noise.Other than that,the Cramér-Rao bound(CRB)under impulse noise environment has been drawn to test the capability of the presented method.
基金supported by the National Natural Science Foundation of China(62073093)the initiation fund for postdoctoral research in Heilongjiang Province(LBH-Q19098)the Natural Science Foundation of Heilongjiang Province(LH2020F017).
文摘In order to solve the problem that the performance of traditional localization methods for mixed near-field sources(NFSs)and far-field sources(FFSs)degrades under impulsive noise,a robust and novel localization method is proposed.After eliminating the impacts of impulsive noise by the weighted out-lier filter,the direction of arrivals(DOAs)of FFSs can be estimated by multiple signal classification(MUSIC)spectral peaks search.Based on the DOAs information of FFSs,the separation of mixed sources can be performed.Finally,the estimation of localizing parameters of NFSs can avoid two-dimension spectral peaks search by decomposing steering vectors.The Cramer-Rao bounds(CRB)for the unbiased estimations of DOA and range under impulsive noise have been drawn.Simulation experiments verify that the proposed method has advantages in probability of successful estimation(PSE)and root mean square error(RMSE)compared with existing localization methods.It can be concluded that the proposed method is effective and reliable in the environment with low generalized signal to noise ratio(GSNR),few snapshots,and strong impulse.
文摘为了解决冲击噪声下长短时记忆(long short term memory,LSTM)神经网络调制信号识别方法抗冲击噪声能力弱和超参数难以确定的问题,本文提出了一种演化长短时记忆神经网络的调制识别方法。利用基于短时傅里叶变换的卷积神经网络(convolution neural network,CNN)去噪模型对数据集去噪;结合量子计算机制和旗鱼优化器(sailfish optimizer,SFO)设计了量子旗鱼算法(quantum sailfish algorithm,QSFA)去演化LSTM神经网络以获得最优的超参数;使用演化长短时记忆神经网络作为分类器进行自动调制信号识别。仿真结果表明,采用所设计的CNN去噪和演化长短时记忆神经网络模型,识别准确率有了大幅度的提高。量子旗鱼算法演化LSTM神经网络模型降低了传统LSTM神经网络容易陷于局部极小值或者过拟合的概率,当混合信噪比为0 dB,所提方法对11种调制信号的平均识别准确率达到90%以上。
