【作者】 隋丹；

【导师】 葛临东；

【作者基本信息】 解放军信息工程大学 ， 信号与信息处理， 2007， 博士

【摘要】 通信信号盲检测是基于软件无线电的智能化盲接收机中一个重要的信号处理功能,特别是在非协作通信中需要在缺乏先验信息的条件下监视和截获空中信号,因此更显得尤为重要。结合实验室承担的军队大型重点工程研究项目,本文围绕通信信号的盲检测技术展开研究,具体包括通信信号的存在性检测技术、参数估计技术以及对特定信号的匹配识别技术。本文的工作可以概括为以下五方面：1.在信号存在性检测方面,重点研究了短突发以及质量较差信号的检测问题。提出一种基于观测数据时域自相关函数波动性的检测算法,利用观测数据自相关函数的一阶波动函数作为检测函数,与所提出的基于排序-做商-求最大值(SDM)的自适应门限确定方法相结合,较好地解决了对持续时间较短以及相邻间隔较近的突发信号的存在性检测。首次将语音信号处理中的谱熵法引入到通信信号检测中,提出一种基于观测数据短时傅氏变换幅度谱熵值的检测算法,实现了低信噪比下的信号检测。2.在信噪比估计方面,重点研究了与调制方式、载波频率、符号速率等参数无关的中频观测数据的盲信噪比估计问题。提出一种基于改进广义分符矩(GSSME)算法的MPSK信号盲信噪比估计算法,利用信噪比估计值的标准差作为算法迭代是否终止的度量,有效解决了原算法很难确定最佳迭代次数的问题。详细推导了基于子空间分解技术的盲信噪比估计算法的估计均值和估计方差的表达式,并且将该算法应用于OdB以下的QAM信号,弱化了原算法的应用条件,拓展了原算法的适用范围。提出一种基于改进投影近似子空间跟踪(PASTd)算法的盲信噪比估计算法,克服了基于子空间分解算法需要对观测数据的自相关矩阵进行特征值分解,运算量较大的问题,同时将Gram-Schmidt正交化过程引入到PASTd中,使计算得到的特征向量相互正交,从而保证算法具有更好的收敛性能。并且该算法的估计结果与观测数据的调制方式、载波频率等参数无关,也不要求系统同步。3.在符号速率估计方面,重点研究了调制方式未知时中频观测数据的盲符号速率估计问题。提出一种综合利用离散小波变换和连续小波变换的CPFSK信号盲符号速率估计算法,对离散小波变换后的细节信号进行连续小波变换,有效减小了噪声的影响,使低信噪比时的估计性能得到明显改善。提出一种基于小波多分辨率分析的盲符号速率估计算法,通过对观测数据进行多级多分辨率分解,利用各级细节系数之间的相关性来估计符号速率。估计结果不受调制方式、载波频率等参数的影响,适用于在调制方式识别前的符号速率估计。4.在目标信号匹配识别方面,重点讨论了单路PSK信号和FSK信号的有效识别问题。提出一种与调制方式无关的匹配识别算法,利用符号速率、调制阶数和频率间隔作为模板参数,通过一个统一的流程较好地实现了对短波PSK信号和FSK信号的有效识别。提出一种差分-门限检测算法,将对FSK信号调制阶数、频率间隔、载波频率等参数的估计问题转化为对其功率谱(或高次方谱)的谱峰搜索问题。5.结合实验室承担的科研任务,设计实现了一个突发信号存在性检测软件平台和一个目标信号匹配识别软件平台,分别针对大量仿真信号和实际信号给出了详实的实验结果,证实了算法的可行性和系统的有效性。更多还原

【Abstract】 The blind detection of the communication signals is a very important function of the intelligent receiver based on Software Defined Radio (SDR), especially in the case of non-cooperative communications where no a priori information can be used for surveillance and interception.This dissertation is devoted to a study of key technologies of the blind detection of the communication signals, including the presence detection, blind parameter estimation and recognition of the specific signals. The achievements presented in this paper are a part of a large scale army project of research undertaken by the lab the author works with. The contributions obtained in this thesis can be summarized in the following five aspects.1. In the aspect of presence detection, the emphasis is laid on the blind detection of signals with short burst and/or with poor quality.A novel presence detection algorithm based on the fluctuation of the correlation function of the received signals in time domain is proposed. The algorithm, employing the first-order fluctuation function of the correlation function as the detection function and combining with the SDM-based adaptive threshold estimation method proposed in this paper, provides an effective solution to the detection problems of short burst or short burst-interval signals.A robust detection algorithm based on the spectral entropy of the short-time Fourier Transform (STFT) of the received signals, which is originally used in the speech signal processing, is proposed for the detection of signals under low Signal-to-Noise Ratio (SNR).2. In the aspect of blind SNR estimation, some algorithms independent of the parameters of the signals such as modulation type, carrier frequency and symbol rate etc. are addressed and discussed.A modified Generalized Split-Symbol Moment Estimator (GSSME) algorithm is proposed for the blind SNR estimation of MPSK signals, and the standard derivation of the estimated SNR is suggested as a measurement for indicating the end of an optimal iteration, which is a great difficulty in the original algorithm.The analytical expressions for the lower bounds of the mean and variance of the estimation are derived in detail for the subspace-based blind SNR estimator. In addition, simulations are performed for MQAM signals besides MPSK and MFSK signals, especially under negative SNR circumstance, expanding the applicable range of the original algorithm.A blind SNR estimator based on the modified Projection Approximation Subspace Tracking deflation (PASTd) algorithm is proposed. The orthogonality of the estimated eigenvectors is guaranteed by introducing the Gram-Schmidt orthogonization process into the original PASTd method. Compared with the eigenvalue decomposition-based method, the proposed algorithm can achieve a more accurate estimation with lower computational complexity. Furthermore, the algorithm needs neither the parameters of the received signals, such as the carrier frequency, symbol rate and modulation scheme, nor the synchronization of the system.3. In the aspect of blind symbol rate estimation, the investigation is targeted at the algorithms for signals with unknown modulation type and nonzero carrier frequency.A novel blind symbol rate estimation algorithm for CPFSK signals is proposed based on combination of Discrete Wavelet Transform (DWT) with Continuous Wavelet Transform (CWT). CWT is used for the extraction of detailed information of the signal obtained from the DWT, and greatly reduces the influence of the noise. Thus the proposed algorithm can get an accurate estimation in low SNR.A blind symbol rate estimator based on the wavelet multi-resolution analysis is proposed. The wavelet orthogonal multi-resolution decomposition is performed for the received signals and the symbol rate can be estimated via the spectrum of the correlation function of the detailed coefficients at different stages. The performance of the proposed algorithm is less influenced by the parameters of the received signals, such as modulation type and carrier frequency, thus is very suitable for the estimation before modulation identification.4. In the aspect of recognition of the target signals, the emphasis is put on the effective identification of single-path MPSK and MFSK signals.A matched recognition alogirthm irrespective of the modulation type of the received signals is proposed. Effective recognition of the short wave PSK and FSK signals can be implemented in a uniform flow chart by employing parameters, including symbol rate, modulation order and frequency interval, as the template parameters.A method for peak-value-search, named as difference-threshold algorithm, is proposed, which converts the parameter estimation of MFSK signals, including the modulation order, frequency interval and carrier frequency, into a peak search problem of the Power Spectral Density (PSD) function or its higher-order counterparts.5. According to the requirements of the tasks the author undertaken in the project, a software platform for presence detection of burst signals and a software platform for matched recognition of target signals are designed and implemented respectively. Detailed experimental results with regard to various simulated signals as well as practical signals are provided which have proved the feasibility and effectiveness of the above two platforms.更多还原