《雷达信号脉内分析与处理理论及算法研究》由会员分享，可在线阅读，更多相关《雷达信号脉内分析与处理理论及算法研究（112页珍藏版）》请在金锄头文库上搜索。
1、南京航空航天大学 博士学位论文 雷达信号脉内分析与处理理论及算法研究 姓名：邓振淼 申请学位级别：博士 专业：信号与信息处理 指导教师：刘渝 20070401 南京航空航天大学博士学位论文 I 摘 要 在电子战中，对敌方雷达的侦察和干扰是非常重要的课题，而这些都离不开对 截获雷达信号的分析。在雷达信号分析中，雷达信号脉内分析算法具有非常重要的 地位。通过脉内分析，可以了解敌方雷达的效能，可以给作战部队提供支援，提供 作战和干扰的手段。本文在前人工作的基础上，对雷达脉内分析的若干关键问题进 行了研究。本文的主要创造性成果有： (1)提出了修正 Rife(MRife)算法。通过对信号进行频移，使新
2、信号的频率位于两 个相邻量化频率点的中心区域，然后再利用 Rife 算法进行频率估计。仿真结果表明 本算法性能不随被估计信号的频率分布而产生波动，整体性能优于牛顿迭代法（一 次迭代） ，接近二次迭代，在低信噪比条件下不存在发散问题，性能比牛顿迭代稳定。 (2)以 Rife 估计为初始值的牛顿迭代存在收敛性问题。针对此问题，本文提出以 修正 Rife 算法为初始值进行牛顿迭代，频率估计值的方差在整个频段都接近克拉美 罗限(CramerRao Lower Bound，CRLB)，具有稳定的性能。 (3)研究了应用于全相位频谱分析时的正弦波频率估计算法。首先分析了全相位 频谱的结构，给出全相位频域插
3、值公式；然后通过频移得到修正频域插值公式。与 现有的全相位频率校正算法比较，频域插值算法具有更高的估计精度，在中低信噪 比时表现出更佳的性能。 (4)提出了基于相关接收机的码速率估计算法和多尺度 Haar 小波变换码速率估计 算法。相关接收机法利用相位编码信号码序列的游程分布特性，对相关接收机输出 的拐点位置进行统计平均得到码速度估计值。Haar 小波变换法以信号带宽作为参考 选取三个小波尺度，利用 Haar 小波的边缘检测特性进行多尺度码元相位跳变点检 测，最后对检测的结果作 DFT 估计频率得到 MPSK 信号的码速率估计。这两种方法 不需要接收信号的先验知识，是真正意义上的盲估计。 (5
4、) 提出了一种相位编码信号的载频盲估计算法。在没有先验知识的情况下，首 先估计接收信号的载频和码速率，然后把信号下变频到基带。利用基带相位编码信 号的相位函数是分段线性函数且每一段直线的斜率均正比于频偏的性质，对相位函 数进行处理，最后运用最小二乘拟合法得到直线的斜率从而得到频偏估计。仿真表 明本算法在没有先验知识的情况下可以对 MPSK信号载频进行精确估计。 雷达信号脉内分析与处理理论及算法研究 II (6)提出了基于最大似然检验的多相码信号调制方式识别算法和多相码信号识别 与参数估计综合算法。基于最大似然检验的调制方式识别算法首先估计接收信号的 载频和码元宽度，接着构造本地参考信号和相关接
5、收机，最后运用最大似然准则对 相关接收机的输出进行判决，选择最大输出所对应的参考信号作为接收信号的调制 方式。调制方式识别与参数估计综合算法把本地参考信号与接收信号共轭相乘，利 用当本地参考信号与接收信号的调制方式一致时共轭相乘的结果是正弦波的性质进 行调制方式识别，最后估计该正弦波的频率作为载频估计。 关键词：脉内分析，频率估计，牛顿迭代，全相位，相位编码，码速率，最小二乘， 多相码，最大似然 南京航空航天大学博士学位论文 III Abstract In the Electronic Warfare, reconnaissance and jamming to the threatening
6、 radar was a very important subject. The analysis of those intercepted radar signals was a must. Intrapulse analysis of radar signals plays a significant role in the radar signal analysis. Through the Intrapulse analysis， the performance of the enemy radar can be evaluated and the support can be pro
7、vided to the operational force, such as the method of attack and defense. This thesis studies several key problems of radar intrapulse analysis. The main contributions are summerized as follows: Firstly, a modified Rife (MRife) algorithm is proposed. It is operated by moving the signal frequency to
8、the central region of two neighboring discrete frequencies and then the frequency is estimated by Rife algorithm. Simulation results show that the performance of MRife does not fluctuate with the distribution of signal frequency, RMSE(root mean square error) is less than that of one iteration of New
9、tons method, and close to that of two iterations. Under the condition of low signal-to-noise ratio (SNR), MRife has no divergence problem and is more stable than Newtons method. Secondly, the divergence problem of Newton Iteration with Rife estimation as its initial value is illustrated. In order to
10、 solve the problem, the starting value of Newtons method is changed to MRife estimation. Through this modification, the RMSE of frequency estimation is close to CRLB throughout the frequency band and the performance is steady. Thirdly, frequency estimation algorithms of sinusoid applied to all phase
11、 spectrum analysis are studied. First the structure of all phase spectrums is analyzed and the all phase spectral interpolation formula is put forward. Then the modified all phase spectral interpolation algorithm is presented by frequency shift. Compared with the all phase frequency correction metho
12、d, the spectral interpolation algorithm is more accurate and achieves better performance under medium and low SNR conditions. Fourthly, a symbol rate estimator based on correlation receiver and blind symbol rate estimation method based on multi-scale Haar wavelet transform are presented. According t
13、o the run length distribution characteristic of multi-phase-shift keying (MPSK) signals, the method based on correlation receiver calculates the assembly average of the position of spinodals, which are the output of correlation receiver. Then the symbol rate can be estimated. For the method of Haar
14、wavelet transform, three wavelet scales are choosed according to signal bandwidth. The phase jumps, which are the output of the receiver, are detected with the multi-scale Haar Wavelet. The symbol rate is extracted by Fourier transform to the detection results. These two algorithms can operate witho
15、ut prior 雷达信号脉内分析与处理理论及算法研究 IV knowledge and realize the real blind estimation. Fifthly, a blind carrier frequency estimation algorithm of MPSK signals is proposed. First the carrier frequency and symbol rate are estimated without prior knowledge. Then the signals are transformed to the baseband. Be
16、cause the phase of baseband-modulated PSK signal is a piecewise linear function and the slope coefficient of every segment is a linear function of frequency offset, the multi-segment line can be transformed to a straight-line and then the least squares fitting method is applied to estimate the slope coefficient. The simulation results show that this method can accurately estimate the carrier frequency without prior knowledge. Sixthly, a maximum-likelihood-based recognition algorithm of modu
