【作者】 张晓林；

【导师】 郭黎利；

【作者基本信息】 哈尔滨工程大学 ， 通信与信息系统， 2007， 博士

【摘要】 直接序列扩频通信目前在军事通信和民用CDMA移动通信中获得了广泛应用。对于合作的直扩信号,要想获得其携带的通信信息,必须事先掌握载波频率,码元速率,伪随机码型等诸多参数。而对于非合作的直扩信号,由于其功率谱密度很低,伪随机码型未知等因素,再加上复杂的环境背景噪声,要想截获十分困难。因此针对直扩信号的检测和侦听一直是研究的热点。可见研究出一种能够有效检测直扩信号同时又能识别其参数的有效方法,对军事通信和民用通信都具有重要意义。在本文中根据背景噪声的不同,提出了几种检测直扩信号的方法,并通过实验仿真验证了算法的可行性与有效性。当直扩信号中仅仅包含单频干扰或窄带干扰时,论文首先提出了一种新的窄带干扰抑制算法-边带相关置换(SCR)算法。SCR算法的应用前提是多数信号频谱具有对称性,直扩信号更是如此。当单频或窄带干扰只处于直扩信号的一个边带时,可以利用与其对称的边带分量置换干扰,然后提取出直扩信号的频谱。仿真实验表明SCR算法能够有效抑制窄带和单频干扰,性能上要优于变换域干扰切除方法。由于SCR算法不能抑制高斯噪声,论文提出了第2种算法-循环边带相关置换(CSCR)算法。CSCR算法的核心思想是将SCR算法应用到循环谱分析中,实现高斯噪声与窄带干扰的联合抑制。CSCR算法利用高斯噪声不具有循环平稳性,通过循环谱分析抑制高斯噪声,然后在直扩信号的谱相关密度中应用SCR算法来实现窄带干扰的有效抑制。当背景噪声中存在多种干扰,如果干扰频率与直扩信号载频相同时,或者干扰带宽与直扩信号带宽接近时,通过SCR与CSCR算法很难提取出直扩信号。本文在第4章提出了一种从多种干扰中提取直扩信号的检测方案,与传统检测方案不同的是该方案引入了独立分量分析(ICA)技术。首先应用ICA技术从各种干扰中分离出直扩信号,然后识别其调制方式与参数。第4章将FastICA算法与JADE算法引入了检测方案中,该方案具有较高的实时性时,信号的还原性也比较好。在实际工作中,到达接收机信号的分布是不同的,既有亚高斯信号,也有超高斯信号,还可能有近高斯信号,因此要求检测方案中ICA算法的鲁棒性要好。在论文最后提出了一种基于核独立分量分析的直扩信号检测方案。仿真实验表明,该方案的鲁棒性要优于基于FastICA和JADE的检测方案。更多还原

【Abstract】 Direct-Sequence Spread Spectrum (DSSS) communication has been widely applied in military communications and public mobile communications. Before capturing the information carried by cooperative direct-sequence (DS) signals, some important parameters, such as carrier frequency, data rate, and the code pattern of the pseudo-noise (PN) code have to be known. It is very difficult to intercept a non-cooperative DS signal because of its low power spectral density, unknown code pattern of the PN code, and the complex background noise. Therefore, detection and interception of DS signals are always research hotspots. It is of great significance in military communications and public mobile communications to find an effective algorithm for detecting DS signals and recognizing their parameters.According to different background noises, in the thesis we propose several approaches for DS signal detection. Computer Simulations have been made to verify the effectivity of these approaches.To detect the DS signals that are affected only by single-frequency interference or narrow band interference (NBI), a sideband correlation replacement (SCR) algorithm has been introduced in Chapter 2. The SCR algorithm is used on the premise that the frequency spectra of most signals have a symmetrical structure, especially DS signals. If single-frequency interference or NBI is within one sideband of a DS signal, the spectrum of the DS signal can then be captured after replacing the interference with another symmetrical sideband. The Simulation results have verified that the SCR algorithm is effective on the suppression of NBI and single-frequency interference, and has a better performance than the transform-domain interference excise algorithm.However, SCR algorithm cannot suppress Gaussian noise. In Chapter 3 cyclic sideband correlation replacement (CSCR) algorithm is proposed to solve this problem. Gaussian noise and NBI can be suppressed together by using CSCR. The key idea of CSCR algorithm is applying SCR algorithm in the analysis of the cyclic spectrum. To be specific, first Gaussian noise can be reduced through analysis of the cyclic spectrum, because it has no cyclostationary. Then, The NBI is suppressed by using SCR algorithm.SCR or CSCR algorithms will fail to capture the DS signal if there are kinds of interference in the background noise, e.g., the interference has the same carrier frequency as DS signal, or the bandwidth of the interference closed to that of DS signal. A detection scheme that extracts DS signal from the complex background noises is investigated in chapter 4. Different from conventional detection schemes, this scheme has made use of independent component analyze (ICA) algorithm, which can separate the DS signal from kinds of interference, and recognize the type of the modulation and the parameters. In Chapter 4, FastICA and JADE algorithms are also introduced to the detection schemes. These schemes are Real-Time and the signals can be well reconstructed.In practice, signals arriving at the receiver, which could be super-Gaussian, sub-Gaussian and near Gaussian, may have different distribution. Therefore, the ICA algorithm is required to be robust. A DS signal detection scheme based on kernel independent component analyze (KICA) has been presented in Chapter 5. The Simulation results verified that this scheme has the better robustness than those based on FastICA and JADE.更多还原