【作者】 王明阳；

【导师】 周一宇；

【作者基本信息】 国防科学技术大学 ， 信息与通信工程， 2006， 博士

【摘要】 本论文所研究的超宽带(UWB)信号均指超宽带冲激无线电(UWB-IR)信号。UWB系统以低占空比的超短冲激脉冲作为载体,无需载波调制,采用跳时扩谱(TH-UWB)或直接序列扩谱(DS-UWB)技术实现信息传输。与传统无线电系统相比,UWB具有更低的功率谱密度和更宽的带宽,具有高数据率、强抗干扰的特点。UWB信号常常被隐蔽在环境噪声和其它干扰信号中,难以检测,是一种很有军事应用前景的低截获/检测概率信号。超宽带无线电技术在高速军事通信、无人机低截获数据链、高分辨率雷达/定位系统中已经得到实际应用。因此在非合作条件下,研究对UWB无线电信号的检测技术,具有重要的军事意义。本文研究了在负信噪比且没有先验信息或仅有部分先验信息条件下, UWB无线电信号的检测问题。论文的主要工作如下:在全盲和负信噪比情况下,研究了基于希尔伯特-黄变换(HHT)的UWB信号检测方法。仿真实验表明:在进行UWB信号检测时,与传统小波分析方法和Wigner-Ville分布相比,希尔伯特-黄方法具有优越性。利用信号和噪声在固有模态函数域中有不同的性态表现,提出了一种固有模态函数积检测器(IMFs product detctor),在低信噪比和背景噪声分布未知条件下,IMF积检测器优于Teager能量算子(TEO)检测器。参考Rosenfeld子带乘积理论,提出了基于固有模态函数域滤波(IMFDF)方法,用于对带噪信号的边缘检测。IMFDF和小波域滤波方法相比,相对简单,计算复杂度小。提出了可用于全盲信号检测的CCIO测度准则和两种基于参数可调非周期随机共振(PASR)的UWB信号检测方法。利用PASR系统输出信号和输入信号最匹配时,随机共振最显著的这一现象,区别于Collins互相关方法,提出了基于输入(信号+噪声)输出互相关的ASR测度方法,即CCIO方法。利用线性响应理论(LRT),推导了CCIO测度方法的性能,验证了CCIO测度下的PASR方法可以实现全盲负信噪比(低于? 10dB)非合作UWB信号检测。在CCIO准则下,提出了基于PASR的DS-UWB系统码序列检测方法。对于低信噪比DS-UWB而言, PASR检测器逊于匹配滤波检测器;但当匹配滤波器存在较大失配时(出现同步误差),结论完全相反。研究发现:PASR方法的检测性能随采样频率的提高而改善。在CCIO准则下,利用多窗并行检测的思想,提出了基于PASR的多时窗检测器(PASR-MWD),用于TH-UWB信号检测。仿真试验表明:在负信噪比情况下,PASR多时窗检测器优于多辐射计组和固有模态函数积检测器,但逊于匹配滤波检测方法。在负信噪比和全盲的条件下,研究了基于分段自相关积累(PACA)的UWB(DS-UWB和TH-UWB)信号检测方法。经过严格数学推导,指出了TH-UWB信号的PACA函数不但在字符周期整数倍处出现离散峰值,而且按三个峰值(一大峰值左右对称出现小峰值)成对出现,大峰和小峰的间隔就是PPM调制时间偏移。仿真验证了通过增加积累时间,实现负信噪比UWB信号检测的可行性。研究了仅需已知字符周期情况下基于特征值分解方法的UWB码序列波形检测方法,首次理论推导了DS-UWB、TH-UWB特征值和码序列波形的分布情况。提出了基于特征值分解的TH-UWB码序列波形估计方法,指出了在时间窗和字符非同步情况下,将会出现三个大的特征值,对应的三个特征向量反映了二元字符跳时序列波形之和的特性。利用矩阵扰动理论,证明了码序列波形估计误差的引理。分析了波形估计误差和输入信噪比、积累窗个数的关系。更多还原

【Abstract】 Ultra-Wide Band (UWB) signals in this dissertation is restricted in UWB impulse radio (UWB-IR) signals.UWB systems rely on ultra-short (nanosecond scale) impulse pulse waveforms of low duty ratio that can be free of sine-wave carriers and transform information via time-hopping(TH) or direct spread(DS) technology. Comparing to traditional radio systems, UWB systems own lower power density of frequence (PDF) and wider bandwith, which can provide extremely high data rate and good property of anti-jamming. For impulse pulse often be covert in environmental noise or the other jamming, UWB exhibits low probability of detection (LPD), which be used in military domain especially. UWB signals are adopted by the military communication of high data rate, low probability of interception (LPI) data link between an unmanned aerial vehicle(UAV) and a ground station, and high resolution radar/location systems.Under the condition of non-cooperative reception, the research on detection technology of UWB signal has important military signification.This dissertion is focused on the detection problems of UWB radio signal under the condtion of unknown or partly known apriori information, while signal-to-noise ratio (SNR) is below 0dB. The main work follows below:While the input SNR is below 0dB and apriori information are all blind, we study the detection methods of UWB signal based on Hilbert-Huang transform (HHT). The conclusion shows that HHT is superior to traditional wavelet transform and Wigner-Ville distribution.For the different property between signal and noise in the intrinsic mode function (IMF) domain, IMFs product detector is proposed. Under the condition of low SNR and the unknown distribution of noise, Simulation shows that IMFs product detector is superior to Teager energy operator (TEO) detector. Refering to Rosenfeld subband product theory, IMF domain filter (IMFDF) is developed and used in the edge detection. In contrary to wavelet transform domain filter, IMFDF is more simple, and has less complexity.The measurement rule of coss correlation of input (signal add noise) and output (CCIO) for blind signal detection is proposed. And then, two methods of UWB signal detection via parameter-induced aperiod stochastic resonance (PASR) are presented.While the input signal matches the output signal of PASR system, phenomena of stochastic resonance is more prominent. So, we proposed CCIO measurement methods, which is different to Collins cross correlation methods.Using linear response theory (LRT), the theoretically performance of CCIO is deduced. While the input SNR is below -10dB and the signal is all unknown, the detector based on PASR via CCIO measurement can work well.Under the rule of CCIO measurement, code sequence detection methods of DS-UWB signal via PASR system are developed. For lower SNR DS-UWB signals, the performance of PASR detector is inferior to matched filter (MF). On the contrary, if there is large error of desychronisation, performance of PASR detector is superior to MF.Under the rule of CCIO measurement, the PASR multiple windows detector (PASR-MWD) is proposed to detect the TH-UWB signals. The results show that the performance of the PASR-MWD is inferior to MF, while superior to multi-radiometer and IMFs product detector when the input SNR is below 0dB.If there is no apriori knowledge and the SNR is below 0dB, the methods of piecewise auto-correlation accumulation (PACA) for the detection of UWB signals is presented. After rigorously mathematical deduction, some valuable conclusions are drawn: at the multiple times of symbol period, discrete peaks of the PACA function appear with 3 peaks in group, which includes a larger peak located in the middle and the other smaller two located symmetricly beside the larger one. The distance between small peaks is modulation offset of pulse position PPM. The simulation shows that enhance the accumulation time, the detection of UWB signal under 0dB is feasible.If the symbol period is known only, the methods of detection UWB code sequence waveform based on the eigenvalues decomposition(EVD) is developed. The distribution of eigenvalue and eigenvectors of the covariance matrix of UWB signal is deduced theoretically. The estimation methods of TH-UWB code sequence based on EVD are proposed especially.The phenomena of appearing 3 eigenvectors corresponding to 3 bigger eigenvalues reflects the sum of 2-arry TH symbol’s waveform in the asynchronous situation for TH-UWB systems. The error lemma of estimation of code sequence via matrix perturbation theory is proved. At last, the relations among the error of estimation, the input SNR, and the number of accumulation windows are analyzed.更多还原