基于卷积神经网络的MIMO空分复用发射天线数目识别

针对传统多输入多输出(Multiple-Input Multiple-Output,MIMO)空分复用发射天线数目识别方法要求对于非合作方接收天线数目大于发射天线数目,以及特征寻找困难的问题,提出了一种基于卷积神经网络的MIMO空分复用发射天线数目识别方法.通过单天线接收目标信号,采集数据,再由卷积神经网络(Convolutional Neural Network,CNN)训练和测试.网络用发射端信号经过空时编码后表现出的不同特征识别发射天线数目.仿真和实测数据验证了CNN对于发射天线数目识别的效果,在信噪比大于15 dB时识别率可以达到85％.     

基于CRNN混合神经网络的多语种识别

在语种识别过程中,为提取语音信号中的空间特 征以及时序特征,从而达到提高多语 种识别准确率的目的,提出了一种利用卷积循环神经网络(convolutional recurrent neural network,CRNN)混合神经网络的多语种识别模型。该模型首先提 取语音信号的声学特征;然后将特征输入到卷积神经网络(convolutional neural network,CNN) 提取低维度的空间特征;再通过空 间金字塔池化层(spatial pyramid pooling layer,SPP layer) 对空间特征进行规整,得到固定长度的一维特征;最后将其输入到循环神经 网络(recurrenrt neural network,CNN) 来判别语种信息。为验证模型的鲁棒性,实验分别在3个数据集上进行,结果表明:相 比于传统的CNN和RNN,CRNN混合神经网络对不同数据集的语种识别 准确率均有提高,其中在8语种数据集中时长为5 s的语音上最为明显,分别提高了 5.3% 和6.1%。     

一种同型空时分组码的识别算法

针对空时分组码(Space-Time Block Code,STBC)盲识别中码型相同的编码区分性较差甚至无法区分的问题,提出了一种基于接收信号统计特征的识别算法.首先分析了多输入多输出(Multiple Input Multiple Output,MIMO)系统中采用的空时编码方案与接收信号的统计特征之间的相关性,设计了概率匹配与弥散度匹配对该相关性进行量化,获得接收信号与不同编码方案的匹配度,最后利用决策树选择匹配度最高的编码作为识别结果.仿真结果表明,针对两组同型的空时分组码,所提算法在信噪比为8 dB时可达98%以上的识别率,而基于特征提取的传统算法无法对两组编码进行有效区分;与基于深度学习的算法相比,本文算法对同型空时码的识别具有更好的鲁棒性,识别过程使用更少的计算资源.     

基于神经网络的移动通信网络异常信号识别优化

常规的移动通信网络异常信号识别是在理想环境下进行的,实际的异常信号识别往往受到其他干扰的影响,出现识别误差的问题。因此,设计了基于神经网络的移动通信网络异常信号识别优化方法。提取移动通信网络异常信号特征,将短时能量信号与过零信号的时域提取出来,过滤时域上的噪声信号,保留异常信号存在的部分。基于神经网络构建通信网络异常信号识别模型,将异常信号特征神经元作为输入,加权求和输入神经元特征,以激活阈值判断当前信号是否异常,从而优化异常信号识别精准度。优化网络异常信号识别的回归损失,降低模型训练损失,从而符合模型输出预期。采用对比实验,验证了该方法的识别准确性更高,优化效果更佳,能够应用于实际生活中。     

残差膨胀卷积结构下的多模态特征调制方式识别

自动调制方式识别技术在通信领域有着不可或缺的作用,针对传统的卷积神经网络在信号分类问题中特征提取能力不足的问题,本文研究了一种利用多维度特征的端到端双流膨胀卷积神经网络来对调制信号进行分类的方法。该方法不仅利用原始采样信号,还利用输入信号的瞬时幅度和相位信息;原始IQ(In-phase and Quadrature, IQ)数据输入进神经网络后,网络首先通过内置的数据预处理模块对输入的IQ信号进行预处理,提取原始信号的幅度和相位信息,再将原始IQ信号和幅度相位两种特征信息分别通过两个并行的卷积神经网络结构分别进行特征提取;本文所设计的双流卷积神经网络模型中的膨胀残差网络分支利用卷积核的膨胀卷积特性,将膨胀卷积与残差网络结构相结合,在网络参数不变的情况下使得卷积核具有更大的感受野,同时也能够更好地结合上下文信息,另一个网络分支是将卷积神经网络与长短期记忆神经网络相串联,然后将两个并行卷积神经网络的输出特征向量进行矩阵相乘达到两种特征信息融合的目的。整个识别过程是基于端到端的,数据预处理模块内嵌到神经网络内部,由神经网络完成对数据的预处理,只需将原始的IQ数据直接送入神经网络即可;仿真实验...     

基于深度卷积神经网络的协作频谱感知方法

针对传统卷积神经网络(CNN)频谱感知方法提取特征能力受限于网络结构简单,增加网络结构又容易出现梯度消失等问题,该文通过在传统卷积神经网络中添加捷径连接,实现输入层恒等映射更深的网络,提出一种基于深度卷积神经网络(DCNN)的协作频谱感知方法。该方法将频谱感知问题转化为图像二分类问题,对正交相移键控(QPSK)信号的协方差矩阵进行归一化灰度处理,并作为深度卷积神经网络的输入,通过残差学习训练深度卷积神经网络模型,提取2维灰度图像的深层特征,将测试数据输入到训练好的模型中,完成基于图像分类的频谱感知。实验结果表明:与传统的频谱感知方法相比,在低信噪比(SNR)下、多用户协作感知时,所提方法具有更高的检测概率和更低的虚警概率。     

基于深度卷积神经网络的协作频谱感知方法

针对传统卷积神经网络(CNN)频谱感知方法提取特征能力受限于网络结构简单,增加网络结构又容易出现梯度消失等问题,该文通过在传统卷积神经网络中添加捷径连接,实现输入层恒等映射更深的网络,提出一种基于深度卷积神经网络(DCNN)的协作频谱感知方法.该方法将频谱感知问题转化为图像二分类问题,对正交相移键控(QPSK)信号的协方差矩阵进行归一化灰度处理,并作为深度卷积神经网络的输入,通过残差学习训练深度卷积神经网络模型,提取2维灰度图像的深层特征,将测试数据输入到训练好的模型中,完成基于图像分类的频谱感知.实验结果表明:与传统的频谱感知方法相比,在低信噪比(SNR)下、多用户协作感知时,所提方法具有更高的检测概率和更低的虚警概率.     

基于双通道复数卷积神经网络的DOA估计算法

针对低信噪比下基于实数卷积神经网络(RV-CNN)的阵列波达方向(DOA)估计方法对接收信号幅相特征提取不充分的问题,引入复数卷积神经网络(CV-CNN)进行DOA估计。为进一步提高分类准确率,构建了一种基于复数卷积神经网络的非对称双通道DOA估计模型(CV-DCNN)。该模型以阵列接收信号的复数协方差矩阵作为输入,分别输入由空洞卷积层组成的第一通道和由标准卷积层组成的第二通道中,其中空洞卷积在不损失角度信息的情况下,增大特征图的感受野。通过复数卷积神经网络(CV-CNN)独有的复数卷积方式提取和融合信号的幅值和相位特征,将双通道提取的特征融合后通过全连接层和sigmoid函数实现角度分类结果输出。实验结果表明,CV-CNN比RV-CNN有更快的收敛速度,在低信噪比和少快拍条件下,CV-CNN比RV-CNN有更高的估计精度,而CV-DCNN比CV-CNN在收敛速度和估计精度上又有了进一步的提升。     

基于信噪比分类网络的调制信号分类识别算法

针对传统降噪算法损伤高信噪比(SNR)信号而造成信号识别准确率下降的问题,该文提出基于卷积神经网络的信噪比分类算法,该算法利用卷积神经网络对信号进行特征提取,用固定K均值(FK-means)算法对提取的特征进行聚类处理,准确分类高低信噪比信号。低信噪比信号采用改进的中值滤波算法降噪,改进的中值滤波算法在传统中值滤波的基础上增加了前后采样窗口的关联性机制,来改善传统中值滤波算法处理连续噪声效果不佳的问题。为充分提取信号的空间特征和时间特征,该文提出卷积神经网络和长短时记忆网络并联的卷积长短时(P-CL)网络,利用卷积神经网络和长短时记忆网络分别提取信号的空间特征与时间特征,并进行特征融合与分类。实验表明,该文提出的调制信号分类模型识别准确率为91%,相比于卷积长短时(CNN-LSTM)网络提高了6%。     

基于BP神经网络的直扩信号扩频码盲识别

针对微弱直扩信号的盲检测与估计问题,在接收方未知发送方扩频序列的前提下 ,提出了一种恢复直接序列扩频（DSSS）信号扩频码的方法。该方法基于反向传播（B P）神经网络,它的输入是接收到的信号,而其期望输出是和输入相同的信号,根据误差反 向传播来有监督地调节神经网络,网络达到收敛时根据第二层权值的符号函数值即可盲估计 出扩频码序列。实验结果表明,即使是在负信噪比情况下,该方法也能得到一个很好的估计 效果。     

The Augmented State Diagram and its Application to Convolutional and Turbo Codes

Convolutional block codes, which are commonly used as constituent codes in turbo code configurations, accept a block of information bits as input rather than a continuous stream of bits. In this paper, we propose a technique for the calculation of the transfer function of convolutional block codes, both punctured and nonpunctured. The novelty of our approach lies in the augmentation of the conventional state diagram, which allows the enumeration of all codeword sequences of a convolutional block code. In the case of a turbo code, we can readily calculate an upper bound to its bit error rate performance if the transfer function of each constituent convolutional block code has been obtained. The bound gives an accurate estimate of the error floor of the turbo code and, consequently, our method provides a useful analytical tool for determining constituent codes or identifying puncturing patterns that improve the bit error rate performance of a turbo code, at high signal-to-noise ratios.     

Space-time codes and concatenated channel codes for wirelesscommunications

Following a brief historical perspective on channel coding, an introduction to space-time block codes is given. The various space-time codes considered are then concatenated with a range of channel codecs, such as convolutional and block-based turbo codes as well as conventional and turbo trellis codes. The associated estimated complexity issues and memory requirements are also considered. These discussions are followed by a performance study of various space-time and channel-coded transceivers. Our aim is first to identify a space-time code/channel code combination constituting a good engineering tradeoff in terms of its effective throughput, bit-error-rate performance, and estimated complexity. Specifically, the issue of bit-to-symbol mapping is addressed in the context of convolutional codes (CCs) and convolutional coding as well as Bose-Chaudhuri-Hocquenghem coding-based turbo codes in conjunction with an attractive unity-rate space-time code and multilevel modulation is detailed. It is concluded that over the nondispersive or narrow-band fading channels, the best performance versus complexity tradeoff is constituted by Alamouti's twin-antenna block space-time code concatenated with turbo convolutional codes. Further comparisons with space-time trellis codes result in similar conclusions     

List Viterbi algorithms for continuous transmission

The conventional list Viterbi algorithm (LVA) produces a list of the L best output sequences over a certain block length in decoding a terminated convolutional code. We show in this paper that the LVA with a sufficiently long list is an optimum maximum-likelihood decoder for the concatenated pair of a convolutional code and a cyclic redundancy check (CRC) block code with error detection. The CRC is used to select the output. New LVAs for continuous transmission are proposed and evaluated, where no termination bits are required for the convolutional code for every CRC block. We also present optimum and suboptimum LVAs for tailbiting convolutional codes. Convolutional codes with Viterbi decoding were proposed for so-called hybrid in band on channel (hybrid IBOC) systems for digital audio broadcasting compatible with the frequency modulation band. For high-quality audio signals, it is beneficial to use error concealment/error mitigation techniques to avoid the worst type of channel errors. This requires a reliable error flag mechanism (error detection feature) in the channel decoder. A CRC on a block of audio information bits provides this mechanism. We demonstrate how the LVA can significantly reduce the flag rate compared to the regular Viterbi algorithm (VA) for the same transmission parameters. At the expense of complexity, a receiver optional LVA can reduce the flag rate by more than an order of magnitude. The difference in audio quality is dramatic. The LVA is backward compatible with a VA     

FSO-MIMO中的自适应多层空时编码

为了根据信道的时变特性来选择合适的空时编码方式,结合正交空时分组码与分层空时码的优点,并借鉴天线分组的多层空时编码原理,提出了一种在自由空间光通信多输入多输出中4×4的自适应多层空时编码方案,并用Monte Carlo法进行仿真研究。结果表明,在一定的信噪比范围内,采用自适应多层空时,编码方案在保证一定误比特率的条件下,能使数据传输速率达到最大化;采用自适应调制方式能更有效地利用资源并提高数据的传输速率。     

Concatenated codes for fading channels based on recursive space-time trellis codes

We propose a class of codes which combine the principles of turbo coding and space-time trellis codes. It is first shown that several classes of space-time codes have an equivalent recursive realization. This fact is then exploited to design serial concatenated coding schemes with an outer code, interleaver, and an inner recursive space-time encoder. Two solutions are proposed in this paper - the use of convolutional outer codes aimed mainly to improve the power efficiency and the use of very high-rate outer codes to obtain significant improvement in power efficiency with a marginal decrease in spectral efficiency. We show that single parity check based turbo product codes are a good candidate for very high-rate outer codes. Finally, we propose an automatic repeat request scheme based on recursive realizations of space-time codes and show that the proposed scheme provides significant reduction in frame error rate.     

基于Faster-RCNN的极验点选式验证码识别

针对传统方法难以识别极验点选式验证码的问题,文中提出一种基于Faster-RCNN目标检测模型和卷积神经网络的识别方法。通过简化的RPN提高Faster-RCNN对于背景图片文本定位的精度,再设计卷积神经网络对文本进行分类识别,并训练Tesseract-OCR识别库对信息提示文本进行识别,实现背景图片文本识别结果与信息提示文本识别结果一一对应,达到识别此类验证码的目的。实验结果表明,该方法识别此类验证码的准确率达到72.4%。     

融合注意力机制的高分辨人脸识别图像重建

针对由于人脸姿势、光照不均、拍摄环境、拍摄设备等内外部因素造成图像分辨率低的问题,提出融合注意力机制的高分辨人脸识别图像重建模型.首先以低分辨率人脸图像对作为两个生成器输入,通过残差块和注意力模块堆叠网络提取人脸特征信息,进而生成高分辨率人脸图像.训练中使用一个鉴别器来监督两个生成器的训练过程.利用Adam算法对鉴别器...     

Convolutional codes I: Algebraic structure

A convolutional encoder is defined as any constant linear sequential circuit. The associated code is the set of all output sequences resulting from any set of input sequences beginning at any time. Encoders are called equivalent if they generate the same code. The invariant factor theorem is used to determine when a convolutional encoder has a feedback-free inverse, and the minimum delay of any inverse. All encoders are shown to be equivalent to minimal encoders, which are feedback-free encoders with feedback-free delay-free inverses, and which can be realized in the conventional manner with as few memory elements as any equivalent encoder, Minimal encoders are shown to be immune to catastrophic error propagation and, in fact, to lead in a certain sense to the shortest decoded error sequences possible per error event. In two appendices, we introduce dual codes and syndromes, and show that a minimal encoder for a dual code has exactly the complexity of the original encoder; we show that systematic encoders with feedback form a canonical class, and compare this class to the minimal class.     

LOW RATE SPACE-TIME TRELLIS CODES INPOWER LIMITED WIRELESSCOMMUNICATION SYSTEMS

Space-time trellis codes can achieve the best tradeoff among bandwidth efficiency, diversity gain, constellation size and trellis complexity. In this paper, some optimum low rate space-time trellis codes are proposed. Performance analysis and simulation show that the low rate space-time trellis codes outperform space-time block codes concatenated with convolutional code at the same bandwidth efficiency, and are more suitable for the power limited wireless communication system.     

扩展卷积码生成矩阵的统一表述

针对在删除卷积码识别过程中缺乏对扩展卷积码先验认知的问题,提出了一种求解母码与扩展卷积码生成矩阵的统一表述方法。通过分析编码器输入输出关系的基本物理意义,先后以( n,1,m)、(n,k,m)作为母码,构建了其与扩展后编码器多项式系数的对应关系模型,归纳和证明了扩展卷积码生成矩阵的统一表述定理。验证结果表明：该定理能够对扩展卷积码生成矩阵实现快速计算,为遍历和重建删除卷积码的删除图样和母码生成矩阵提供方便。