基于核协同表示与鉴别投影的辐射源调制识别

针对辐射源识别中的特征稳定性不高和低信噪比环境适应性不足等问题,提出了一种基于二次时频分布、核协同表示与鉴别投影的识别方法.首先,通过时频变换、稀疏域降噪和二次特征提取的预处理算法降低噪声干扰和特征冗余,以获取高稳定性的二次时频分布特征;然后,采用核协同表示和鉴别投影思想进行降维学习和字典学习,以提升数据低维表征和类间鉴别能力;最后,通过离线训练完成系统优化并用于分类验证.仿真结果表明,二次时频分布特征具备较高稳定性,识别方法具备较强鲁棒性、时效性和适应性;当信噪比为-10dB时,该方法对8类辐射源信号的整体平均识别率达到96.88%.     

基于栈式自编码机和模糊函数主脊的雷达辐射源信号识别

针对人工提取雷达辐射源信号特征存在提取周期长、特征描述不完备等局限性,提出了一种基于深度学习栈式自编码机和模糊函数主脊的雷达信号识别方法.该方法根据信号模糊函数主脊包含丰富的内在调制信息的特点,从信号中提取用于分类识别的抽象特征.通过对六种雷达辐射源信号进行实验,并对比人工特征提取及其他深度学习方法,结果表明,本文所提方法在信噪比（signal-noise ratio,SNR）为2 dB以上时均能保持100%的识别准确率,SNR为-6 dB时识别准确率仍能保持82.83%以上,明显高于其他方法.即使在包含相同调制类型不同参数的信号环境中,当SNR大于0 dB时识别率均稳定在95.0%以上,SNR降低到-4 dB时识别率也能达到79.0%.证明该方法能有效提取到信号的深层特征,且具有良好的抗噪性能,基本满足实际战场的需求.     

基于CEEMD的压缩感知降噪在雷达信号识别中的应用研究

面对日益复杂的电磁对抗环境,如何有效从复杂噪声背景下提取雷达辐射源信号,成为目前热点难题。为此,提出了基于CEEMD的压缩感知理论降噪方法。原始信号经CEEMD处理,对分解得到的本征模态分量通过计算排列熵确定噪声分量和信号分量,在对噪声分量进行有用信号提取时,采用正交匹配追踪算法替代传统阈值处理方法以恢复信号稀疏性,达到降噪目的,该方法无需信号先验知识,克服了传统阈值选取主要依靠主观判断的问题。通过仿真信号分析和雷达辐射源信号识别应用研究,结果表明,所提方法较CEEMD传统阈值方法对低信噪比条件下雷达辐射源信号识别率有了较大的提升,而且减少了有用细节流失。     

一种噪声环境下的雷达目标高分辨率距离像鲁棒识别方法

随着深度学习技术被应用于雷达目标识别领域,其自动提取目标特征的特性大大提高了识别的准确率和鲁棒性,但噪声环境下的鲁棒性有待进一步研究。该文提出了一种在噪声环境下基于卷积神经网络(CNN)的雷达高分辨率距离像(HRRP)数据识别方法,通过增强训练集和使用残差块、inception结构和降噪自编码层增强网络结构,实现了在较宽信噪比范围下的较高识别率,其中在信噪比为0 dB的瑞利噪声条件下,识别率达到96.14%,并分析了网络结构和噪声类型对结果的影响。      

多网络联合识别辐射源个体的优化方法

针对辐射源指纹特征间差异细微且受噪声干扰容易导致识别率下降的问题,提出了一种基于stacking方法集成多个异构网络识别结果的辐射源个体识别优化算法。利用不同网络结构在低信噪比条件下提取指纹特征的差异性,多个异构网络集成各自的预测结果能够提升对指纹特征的提取能力。同时为避免分类准确率提高造成模型规模过大,本文使用网络规模小且结构差异较大的EfficientNets系列异构网络作为基础网络。实验首先在高斯信道条件下验证了基础网络能够有效识别功率放大器杂散噪声,之后利用stacking等优化算法改进模型整体的性能。结果表明,本方法能够进一步利用信号指纹特征之间差异,与其他方法相比对辐射源个体有更高的识别率。      

基于深度学习的动态数据增量式挖掘方法

为了降低数据挖掘时间,提升数据挖掘精度,提出基于深度学习的动态数据增量式挖掘方法.利用简单处理单元组合构成并行分布式处理器,采用栈式稀疏降噪自编码网络作为动态数据特征提取的深度学习模型,通过逐层贪婪无监督策略完成预训练,使用随机梯度下降优化网络权值,获得动态数据规律.利用粗糙集求解近似动态约简容错偏差与置信区间,对动态...     

基于Welch功率谱和卷积神经网络的通信辐射源个体识别

针对低信噪比条件下通信辐射源个体识别率低的问题,提出了一种基于Welch功率谱和卷积神经网络的通信辐射源个体识别方法.构建了由20个基于ZigBee协议的物联网设备组成的测试平台,将ZigBee信号前同步码部分的Welch功率谱数据作为辐射源指纹特征送入卷积神经网络进行分类.该方法在低信噪比条件下很好地保留了辐射源的指纹特征,结合卷积神经网络强大的微特征提取能力,对辐射源进行了有效分类.实验结果证明,在瑞利信道及低信噪比条件下,所提方法的识别效果明显优于其他方法.     

基于领域自适应的动态噪声辐射源个体识别

现有基于深度神经网络的辐射源识别算法受训练场景限制,当待测信号与训练数据集的信道环境噪声不一致时,网络的识别性能严重退化。为了克服该问题,本文提出一种基于迁移学习的辐射源个体识别算法。该算法结合领域自适应的思想,建立优化模型将不同信噪比下信号的特征对齐,使在特定信噪比下训练的神经网络学习到与信道噪声无关的射频指纹特征,实现对其他信噪比下信号的高准确率识别。仿真实验结果表明,提出的算法显著提升了基于深度神经网络的辐射源个体识别算法在动态噪声条件下的准确率,在待识别信号信噪比下降4 dB的情况下,准确率提升了45.18%。      

基于改进小波降噪的雷达一维距离像识别

针对低信噪比下雷达目标一维距离像质量不高、影响目标识别率的问题,将小波阈值降噪的方法应用到雷达目标一维距离像识别研究中,设计了一种新的小波阈值函数,提出了基于小波阈值降噪的雷达一维距离像识别的方法。利用仿真数据进行实验验证,以LVQ（Learning Vector Quantization）神经网络作为分类识别器,进行目标的分类识别研究。结果表明,将小波阈值降噪用于雷达目标一维距离像识别,在低信噪比时能够有效地降低噪声,提高距离像的质量,从而提高目标一维距离像识别率,同时实验也验证了所提出的新阈值函数相较于原阈值函数能更加有效地降低噪声,提高识别率。     

基于栈式稀疏自编码器的低信噪比下低截获概率雷达信号调制类型识别

针对低截获概率(LPI)雷达信号识别率低且特征提取困难的问题,该文提出一种基于Choi-Williams分布(CWD)和栈式稀疏自编码器(sSAE)的自动分类识别系统。该系统从反映信号本质特征的时频图像入手,首先对LPI雷达信号进行CWD时频分析,获取2维时频图像;然后对得到的时频原始图像进行预处理,并把预处理后的图像送入多层稀疏自编码器(SAE)进行离线训练;最后把SAE自动提取的特征输入softmax分类器,实现雷达信号的在线分类识别。仿真结果表明,信噪比为时,该系统对8种LPI雷达信号(LFM, BPSK, Costas, Frank和T1~T4)的整体平均识别率达到96.4%,在低信噪比条件下明显优于人工设计提取信号特征的识别方法。     

基于扩张残差网络的雷达辐射源信号识别

针对低信噪比条件下,复杂多类雷达辐射源信号识别存在特征提取困难,识别正确率低的问题,本文提出了一种基于时频分析和扩张残差网络的辐射源信号自动识别方法.首先通过时频分析将信号时域波形转换成二维时频图像以反映信号本质特征;然后进行时频图像预处理以保留时频图像完备信息,适应深度学习模型输入;最后构建扩张残差网络以自动提取信号时频图像特征,实现雷达辐射源信号分类识别.实验结果表明,信噪比为-6dB时,该方法对16类雷达辐射源信号的整体识别正确率能够达到98.2%,对时频图像特征相似的类LFM（Linear Frequency Modulation）信号的整体识别正确率超过95%.本文提供了一种新的雷达辐射源信号智能识别方法,具有较好的工程应用前景.     

基于循环谱和稀疏表示的多信号调制识别*1

针对难以识别时频重叠的多信号问题,提出一种不用分离混合信号即可识别信号类型的新方法。该方法针对各种调制循环谱的不同,用稀疏表示提取各信号的特征,最后根据提取的特征利用支持向量机对信号进行识别分类。经理论推导和仿真实验得出:该方法对噪声具有一定的鲁棒性,在较低信噪比条件下仍能保持较好的识别性能,在信噪比为-4 d B时,对单信号和混合信号的正确识别率分别可达到93.5%和90.67%。     

基于去噪卷积神经网络的雷达信号调制类型识别

针对低截获概率雷达(LPI)信号处理复杂,低信噪比条件下识别率低的问题,该文提出一种基于去噪卷积神经网络和Inception网络的信号分类识别系统。首先对8种LPI雷达信号进行Choi-Williams分布(CWD)时频变换,得到2维时频图像,然后使用去噪卷积神经网络进行时频图像去噪处理,最后将图像发送到Inception-V4网络进行特征提取,并使用softmax分类器进行分类,实现LPI雷达信号的有效分类识别。仿真结果表明,该方法在–10 dB信噪比(SNR)下,识别率仍然可以达到90%以上。     

Denoising recovery for compressive sensing based on selective measure

In order to reduce the effect of noise folding (NF) phenomenon on the performance of sparse signal recon-struction,a new denoising recovery algorithm based on selective measure was proposed.Firstly,the NF phenomenon in compressive sensing (CS) was explained in theory.Secondly,a new statistic based on compressive measurement data was proposed,and its probability density function (PDF) was deduced and analyzed.Then a noise filter matrix was constructed based on the PDF to guide the optimization of measurement matrix.The optimized measurement matrix can selectively sense the sparse signal and suppress the noise to improve the SNR of the measurement data,resulting in the improvement of sparse reconstruction performance.Finally,it was pointed out that increasing the measurement times can further enhance the performance of denoising reconstruction.Simulation results show that the proposed denoising recon-struction algorithm has a better improvement in the performance of reconstruction of noisy signal,especially under low SNR.     

扩展稀疏表示稳健HRRP目标特征提取方法

为提高低信噪比下高分辨一维距离像目标识别性能，提出扩展稀疏表示的噪声稳健目标特征提取方法。本方法通过对稀疏表示的扩展，实现对目标高分辨一维距离像局部特征与全局特征的提取。其中，在训练阶段利用支持向量理论与字典学习原理，对特征提取字典进行优化提高特征向量的可分性。在测试阶段，利用因子分析模型匹配方法对去噪声字典进行优化，从而实现对噪声的有效抑制，保证了目标识别系统的噪声稳健性。利用实测数据对本方法性能进行测试，结果表明本方法可在低信噪比条件下有效地恢复目标高分辨一维距离像，并实现较高的识别正确率。     

一种基于时频原子特征的雷达辐射源信号识别方法

提出了一种全新的基于时频原子特征的雷达辐射源信号识别方法.训练阶段,在过完备时频原子库的基础上,以类区分度为度量,提取少数最能区分不同类别信号的时频原子作为一组固定的特征;识别阶段,以原子和信号的内积的绝对值作为分类器的输入特征,采用有监督模糊自适应共振网络进行辐射源的自动识别.对5类典型雷达辐射源信号的实验结果表明,...     

Detection of loop closure in visual SLAM:a stacked assorted auto-encoder based approach

The current mainstream methods of loop closure detection in visual simultaneous localization and mapping (SLAM) are based on bag-of-words (BoW). However, traditional BoW-based approaches are strongly affected by changes in the appearance of the scene, which leads to poor robustness and low precision. In order to improve the precision and robustness of loop closure detection, a novel approach based on stacked assorted auto-encoder (SAAE) is proposed. The traditional stacked auto-encoder is made up of multiple layers of the same autoencoder. Compared with the visual BoW model, although it can better extract the features of the scene image, the output feature dimension is high. The proposed SAAE is composed of multiple layers of denoising auto-encoder, convolutional auto-encoder and sparse auto-encoder, it uses denoising auto-encoder to improve the robustness of image features, convolutional auto-encoder to preserve the spatial information of the image, and sparse auto-encoder to reduce the dimensionality of image features. It is capable of extracting low to high dimensional features of the scene image and preserving the spatial local characteristics of the image, which makes the output features more robust. The performance of SAAE is evaluated by a comparison study using data from new college dataset and city centre dataset. The methodology proposed in this paper can effectively improve the precision and robustness of loop closure detection in visual SLAM.     

Adaptive score level fusion of fingerprint and voice combining wavelets and separability measures

This paper presents an adaptive combinational approach for the score level fusion of fingerprint and voice biometrics, whose performance under adverse noise conditions are investigated systematically. An efficient preprocessing on the raw vector of scores using normalization and wavelet denoising is proposed, to improve the performance of the multibiometric system. The class as well as the score separability measures, under various noise conditions are estimated and combined algebraically, to determine the best integration weights, for the complementary modalities employed. The z-score normalized impostor scores are modelled as white Gaussian noise so that the wavelet denoising techniques can be readily applied. The inter/intra class separability measures from the feature space and the d-prime separability measures from the match score space are estimated in the training/validation phase. The performance of the proposed method is compared with the baseline techniques on score level fusion. Experimental evaluations show that the proposed method improves the recognition accuracy and reduces the false acceptance rate (FAR) and false rejection rate (FRR) over the baseline systems, under various signal-to-noise ratio (SNR) conditions. The proposed biometric solutions will be extremely useful in applications where there are less number of available training samples.     

Digital modulation recognition based on instantaneous information

A new digital modulation recognition algorithm based on the instantaneous information is proposed to improve the recognition success rate in the low signal noise ratio(SNR).First denoising of the instantaneous information is optimized by wavelet filter,which can improve the recognition ability at low SNR.Besides the existing 3 key feature parameters,3 new key feature parameters are proposed to be used as the decision criteria for identifying different types of digital modulation,which simplifies the recognition process and improves the recognition ability at low SNR.The simulations demonstrate that all modulation types of interest have been classified with success rate of no lower than 99% when SNR is 10dB.Even if the SNR is lower than 5 dB,the success rate is over 95.4% for most of the modulation types.