跳频体制下基于盲分离的抗干扰技术研究
发布时间：2019-06-18 09:11
【摘要】：时代在进步,科技在发展,电子战逐渐成为一种重要的现代化作战手段。全世界都在积极地研究各种电子战算法,因此未来的电磁环境将会越来越复杂。常用的跳频通信技术是以牺牲有限频谱资源来提高其抗干扰能力,而这种牺牲也是有限的,因此需要新的方法来改善跳频系统的抗干扰性能。本文利用盲源分离方法的优势,研究了盲源分离方法与跳频通信技术相结合的抗干扰技术,来提高系统的抗干扰能力。主要内容为:研究了基于空间时频分布矩阵的跳频信号盲分离技术。跳频信号是典型的非平稳信号,且不同跳频信号具有不同的跳频图案。本文充分利用跳频信号的这些特征,采用了根据观测信号的平滑伪Wigner-Ville分布来构造一个四维矩阵,检测出自源时频点后,利用特征矩阵近似联合对角化算法估计出酉矩阵,进而分离出源信号。本文在已有的时频点选择方法基础上,对已有的选择方法进行了改进。改进后的基于空间时频分布矩阵的盲分离方法提高了分离性能,但是其算法的复杂度有所增加。研究了基于负熵最大化的FastICA分离算法。在接收端接收的信号都是来自不同的电子设备,所以可以认为观测信号中包含的跳频信号是统计独立的。本文充分利用跳频信号的相互独立性、源信号中最多只有一路高斯信号和高斯信号的负熵最大这些性质,采用了基于负熵最大化的FastICA算法来分离包含多路跳频信号的观测信号。本文在原FastICA算法的基础上,对迭代算法进行了改进。改进后的FastICA算法提高了分离精度,而且其复杂度远远小于基于空间时频分布矩阵的盲分离算法,改进的FastICA算法分离精度随信噪比递增,随干信比增加而有所下降。为了分析盲源分离算法对误比特率性能的影响,本文比较了先盲源分离后解跳与直接解跳的误比特率,前者的误比特率性能要优于后者。研究了跳频信号的欠定混合矩阵估计。虽然跳频信号在时域不是稀疏信号,但是通过短时傅里叶变换后,跳频信号在时频域内具有稀疏性。本文利用跳频信号这一特性,通过K-means聚类算法估计出了混合矩阵。本文提出了一种新的初始聚类中心选择方法,通过比较已有的选择方法,本文提出的新方法对混合矩阵的估计精度要高于其他三种选择方法。盲源分离技术与跳频技术的结合提高了其抗干扰能力,这项研究将会有力推动军事通信抗干扰技术的不断发展,具有重要的军事意义和重大的现实意义。
[Abstract]:In the times of progress, science and technology are developing, and electronic warfare is becoming an important means of modern warfare. The world is actively studying various electronic warfare algorithms, so the future electromagnetic environment will become more and more complex. The commonly used frequency-hopping communication technology is to improve the anti-interference ability of the frequency-hopping system by sacrificing the limited spectrum resources, and the sacrifice is also limited, so a new method is needed to improve the anti-interference performance of the frequency-hopping system. This paper makes use of the advantages of blind source separation method, and studies the anti-interference technology of blind source separation method and frequency-hopping communication technology to improve the anti-interference ability of the system. The main content of this paper is to study the blind separation technology of frequency-hopping signal based on spatial time-frequency distribution matrix. The frequency hopping signal is a typical non-stationary signal and the different frequency hopping signals have different frequency hopping patterns. In this paper, the characteristics of frequency-hopping signals are fully utilized, a four-dimensional matrix is constructed based on the smooth pseudo-Wigner-Ville distribution of the observed signals. After the source frequency points are detected, the matrix is estimated by using the characteristic matrix approximation and the joint diagonalization algorithm, and then the source signals are separated. On the basis of the existing time-point selection method, the existing selection method is improved. The improved blind separation method based on the spatial time-frequency distribution matrix improves the separation performance, but the complexity of the algorithm is increased. The FastICA separation algorithm based on the maximum negative entropy is studied. The signals received at the receiving end are all from different electronic devices, so that the frequency hopping signals contained in the observation signals can be considered to be statistically independent. This paper makes full use of the mutual independence of frequency-hopping signals, the most only one-way Gaussian signal in the source signal and the maximum of the negative entropy of the Gaussian signal, and adopts the FastICA algorithm based on the maximization of the negative entropy to separate the observation signals containing the multi-channel frequency-hopping signals. In this paper, the iterative algorithm is improved on the basis of the original FastICA algorithm. The improved FastICA algorithm improves the separation precision, and the complexity is much smaller than the blind separation algorithm based on the spatial time-frequency distribution matrix. In order to analyze the effect of the blind source separation algorithm on the bit error rate performance, the bit error rate of the first blind source separation and the direct solution jump is compared, and the error bit rate performance of the former is better than the latter. In this paper, the under-determined mixed matrix estimation of frequency-hopping signals is studied. Although the frequency-hopping signal is not a sparse signal in the time domain, the frequency-hopping signal has a sparsity in the time domain by a short-time Fourier transform. In this paper, using the characteristic of frequency-hopping signal, the hybrid matrix is estimated by the K-means clustering algorithm. In this paper, a new method of initial cluster center selection is proposed. By comparing the existing selection method, the estimation accuracy of the hybrid matrix is higher than that of the other three selection methods. The combination of blind source separation technology and frequency-hopping technology has improved its anti-interference ability. The research will push forward the development of the anti-interference technology of military communication, which is of great military significance and great practical significance.
【学位授予单位】：电子科技大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TN973
本文编号：2501358
[Abstract]:In the times of progress, science and technology are developing, and electronic warfare is becoming an important means of modern warfare. The world is actively studying various electronic warfare algorithms, so the future electromagnetic environment will become more and more complex. The commonly used frequency-hopping communication technology is to improve the anti-interference ability of the frequency-hopping system by sacrificing the limited spectrum resources, and the sacrifice is also limited, so a new method is needed to improve the anti-interference performance of the frequency-hopping system. This paper makes use of the advantages of blind source separation method, and studies the anti-interference technology of blind source separation method and frequency-hopping communication technology to improve the anti-interference ability of the system. The main content of this paper is to study the blind separation technology of frequency-hopping signal based on spatial time-frequency distribution matrix. The frequency hopping signal is a typical non-stationary signal and the different frequency hopping signals have different frequency hopping patterns. In this paper, the characteristics of frequency-hopping signals are fully utilized, a four-dimensional matrix is constructed based on the smooth pseudo-Wigner-Ville distribution of the observed signals. After the source frequency points are detected, the matrix is estimated by using the characteristic matrix approximation and the joint diagonalization algorithm, and then the source signals are separated. On the basis of the existing time-point selection method, the existing selection method is improved. The improved blind separation method based on the spatial time-frequency distribution matrix improves the separation performance, but the complexity of the algorithm is increased. The FastICA separation algorithm based on the maximum negative entropy is studied. The signals received at the receiving end are all from different electronic devices, so that the frequency hopping signals contained in the observation signals can be considered to be statistically independent. This paper makes full use of the mutual independence of frequency-hopping signals, the most only one-way Gaussian signal in the source signal and the maximum of the negative entropy of the Gaussian signal, and adopts the FastICA algorithm based on the maximization of the negative entropy to separate the observation signals containing the multi-channel frequency-hopping signals. In this paper, the iterative algorithm is improved on the basis of the original FastICA algorithm. The improved FastICA algorithm improves the separation precision, and the complexity is much smaller than the blind separation algorithm based on the spatial time-frequency distribution matrix. In order to analyze the effect of the blind source separation algorithm on the bit error rate performance, the bit error rate of the first blind source separation and the direct solution jump is compared, and the error bit rate performance of the former is better than the latter. In this paper, the under-determined mixed matrix estimation of frequency-hopping signals is studied. Although the frequency-hopping signal is not a sparse signal in the time domain, the frequency-hopping signal has a sparsity in the time domain by a short-time Fourier transform. In this paper, using the characteristic of frequency-hopping signal, the hybrid matrix is estimated by the K-means clustering algorithm. In this paper, a new method of initial cluster center selection is proposed. By comparing the existing selection method, the estimation accuracy of the hybrid matrix is higher than that of the other three selection methods. The combination of blind source separation technology and frequency-hopping technology has improved its anti-interference ability. The research will push forward the development of the anti-interference technology of military communication, which is of great military significance and great practical significance.
【学位授予单位】：电子科技大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TN973
【参考文献】
相关期刊论文 前2条
1 谭北海;谢胜利;;基于源信号数目估计的欠定盲分离[J];电子与信息学报;2008年04期
2 马明;沈越泓;牛英涛;孔昭煜;;基于空间时频分布的非平稳信号盲分离算法性能研究[J];探测与控制学报;2007年03期
，本文编号：2501358
本文链接：https://www.wllwen.com/kejilunwen/wltx/2501358.html
