确定性盲信道辨识算法研究
发布时间：2018-10-17 08:19
【摘要】：无线数字通信中,受多径传输等信道因素的影响,接收信号会存在码间干扰,因此需要进行信道辨识以对其引起的信号畸变进行校正。无需或仅需少量已知符号的盲信道辨识技术得到广泛关注,基于二阶统计量(Second-Order Statistics,SOS)的盲信道辨识算法所需数据量小、算法复杂度低,是近年来盲信道辨识领域的研究热点。本文以国防某重大科研项目为依托,深入研究更适于实际应用的SOS确定性盲辨识算法。全文主要工作与成果如下:1.结合SIMO信道模型,本文归纳总结出具有普适性的SOS盲辨识算法可辨识条件,即信道条件、信源条件和数据条件。深入研究信道零点分布与SOS确定性盲辨识算法的可辨识性关系,并进行了相关的理论分析与仿真实验。2.针对SOS确定性全盲辨识算法对信道阶数的估计精度依赖性较高这一特点,本文提出一种基于样本排序的信道阶数估计改进算法,算法通过建立与样本数据对应的连续次序对二维空间,利用阶数欠估计时的特定图形结构估计信道阶数,改进算法有效地提升了原算法在较低信噪比下的估计性能。然后提出一种联合信道辨识与均衡的阶数估计算法,算法首先构造具有凸形结构的辨识代价函数,而后提出新颖的加权最小二乘均衡准则并给出相关的理论分析,联合同样具有凸形结构的辨识代价函数与均衡代价函数,在达到算法全局最优解时完成对信道阶数的估计,仿真实验表明:该算法在不同信道条件下的估计性能明显优于现有的其他阶数估计算法,且性能稳定可靠。3.针对SOS确定性全盲辨识算法对信道阶数误差鲁棒性差的问题,本文首先对信道零点分布与阶数过估计之间的联系进行深入的理论分析,发现并证明了由阶数过估计额外引入的“公零点”具有单位圆聚集特性,利用这一特性提出一种基于信道零点分布的盲辨识算法,该算法简单实用且适用范围广。同时,将“公零点”的单位圆聚集特性与具有较低复杂度的改进CR算法相结合,在频域范围内求解信道响应以提高算法在小样本数据条件下的辨识性能,提出一种采用FFT方法的抗阶数过估计盲辨识算法,仿真实验表明:该算法具有较强的信道阶数误差鲁棒性。4.针对全盲辨识算法无法辨识含公零点信道且对信道阶数误差敏感的问题,本文提出一种采用奇异值分解方法的半盲辨识算法,算法通过奇异值分解将信道矩阵分解为两个矩阵乘积的形式,分别利用接收数据和已知符号实现信道辨识,仿真实验验证了所提算法的有效性。然后提出一种基于信道相关性的半盲辨识算法,算法利用接收数据构造的相关矩阵与信道向量的正交关系建立约束方程,并利用少量已知符号以及改进的最小二乘准则建立额外的约束,最终通过最小二乘法得到信道响应的闭式解,该算法性能稳健且辨识精度高,对信道噪声及信道阶数误差均具有较强的鲁棒性。
[Abstract]:In wireless digital communication, due to the influence of channel factors such as multipath transmission, the received signal will have inter-symbol interference (ISI), so channel identification is needed to correct the signal distortion caused by it. Blind channel identification without or only a small number of known symbols has received extensive attention. Blind channel identification algorithm based on second-order statistics (Second-Order Statistics,SOS) requires a small amount of data and has a low complexity. It is a hot topic in the field of blind channel identification in recent years. In this paper, based on a major research project of national defense, the SOS deterministic blind identification algorithm, which is more suitable for practical application, is studied in depth. The main work and results are as follows: 1. Combined with the SIMO channel model, this paper summarizes the identifiable conditions of SOS blind identification algorithm with universality, that is, channel condition, source condition and data condition. The relationship between channel zero distribution and the identifiability of SOS deterministic blind identification algorithm is studied, and relevant theoretical analysis and simulation experiments are carried out. 2. In view of the high accuracy dependence on channel order estimation of SOS deterministic all-blind identification algorithm, an improved channel order estimation algorithm based on sample ordering is proposed in this paper. By establishing the continuous sequence pair space corresponding to the sample data and using the special graph structure of order underestimation to estimate the channel order, the improved algorithm can effectively improve the estimation performance of the original algorithm under lower SNR. Then, an order estimation algorithm for joint channel identification and equalization is proposed. Firstly, the cost function with convex structure is constructed, and then a novel weighted least square equalization criterion is proposed and the related theoretical analysis is given. When the identification cost function and equalization cost function with convex structure are combined, the channel order is estimated when the global optimal solution of the algorithm is reached. The simulation results show that the performance of the proposed algorithm is better than that of other order estimation algorithms under different channel conditions, and the performance is stable and reliable. Aiming at the problem of poor robustness of SOS deterministic all-blind identification algorithm to channel order error, the relationship between channel zero distribution and order overestimation is analyzed in this paper. It is found and proved that the "common zero" introduced by order overestimation has the characteristic of unit circle aggregation. By using this property, a blind identification algorithm based on channel zero distribution is proposed. The algorithm is simple and practical and has a wide range of applications. At the same time, combining the unit circle aggregation of "common zero" with the improved CR algorithm with low complexity, the channel response is solved in the frequency domain to improve the identification performance of the algorithm under the condition of small sample data. A blind identification algorithm against order overestimation using FFT method is proposed. The simulation results show that the algorithm is robust to channel order error. 4. Aiming at the problem that the all-blind identification algorithm can not identify the channel with common zero point and is sensitive to channel order error, a semi-blind identification algorithm based on singular value decomposition (SVD) is proposed in this paper. The channel matrix is decomposed into the product of two matrices by singular value decomposition, and the channel identification is realized by using received data and known symbols, respectively. The simulation results show that the proposed algorithm is effective. Then a semi-blind identification algorithm based on channel correlation is proposed. The constraint equation is established by using the orthogonal relation between the correlation matrix constructed by the received data and the channel vector. Using a small number of known symbols and the improved least square criterion to establish additional constraints, the closed-form solution of the channel response is obtained by the least square method. The algorithm is robust in performance and high in identification accuracy. It is robust to channel noise and channel order error.
【学位授予单位】：解放军信息工程大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TN92
本文编号：2276066
[Abstract]:In wireless digital communication, due to the influence of channel factors such as multipath transmission, the received signal will have inter-symbol interference (ISI), so channel identification is needed to correct the signal distortion caused by it. Blind channel identification without or only a small number of known symbols has received extensive attention. Blind channel identification algorithm based on second-order statistics (Second-Order Statistics,SOS) requires a small amount of data and has a low complexity. It is a hot topic in the field of blind channel identification in recent years. In this paper, based on a major research project of national defense, the SOS deterministic blind identification algorithm, which is more suitable for practical application, is studied in depth. The main work and results are as follows: 1. Combined with the SIMO channel model, this paper summarizes the identifiable conditions of SOS blind identification algorithm with universality, that is, channel condition, source condition and data condition. The relationship between channel zero distribution and the identifiability of SOS deterministic blind identification algorithm is studied, and relevant theoretical analysis and simulation experiments are carried out. 2. In view of the high accuracy dependence on channel order estimation of SOS deterministic all-blind identification algorithm, an improved channel order estimation algorithm based on sample ordering is proposed in this paper. By establishing the continuous sequence pair space corresponding to the sample data and using the special graph structure of order underestimation to estimate the channel order, the improved algorithm can effectively improve the estimation performance of the original algorithm under lower SNR. Then, an order estimation algorithm for joint channel identification and equalization is proposed. Firstly, the cost function with convex structure is constructed, and then a novel weighted least square equalization criterion is proposed and the related theoretical analysis is given. When the identification cost function and equalization cost function with convex structure are combined, the channel order is estimated when the global optimal solution of the algorithm is reached. The simulation results show that the performance of the proposed algorithm is better than that of other order estimation algorithms under different channel conditions, and the performance is stable and reliable. Aiming at the problem of poor robustness of SOS deterministic all-blind identification algorithm to channel order error, the relationship between channel zero distribution and order overestimation is analyzed in this paper. It is found and proved that the "common zero" introduced by order overestimation has the characteristic of unit circle aggregation. By using this property, a blind identification algorithm based on channel zero distribution is proposed. The algorithm is simple and practical and has a wide range of applications. At the same time, combining the unit circle aggregation of "common zero" with the improved CR algorithm with low complexity, the channel response is solved in the frequency domain to improve the identification performance of the algorithm under the condition of small sample data. A blind identification algorithm against order overestimation using FFT method is proposed. The simulation results show that the algorithm is robust to channel order error. 4. Aiming at the problem that the all-blind identification algorithm can not identify the channel with common zero point and is sensitive to channel order error, a semi-blind identification algorithm based on singular value decomposition (SVD) is proposed in this paper. The channel matrix is decomposed into the product of two matrices by singular value decomposition, and the channel identification is realized by using received data and known symbols, respectively. The simulation results show that the proposed algorithm is effective. Then a semi-blind identification algorithm based on channel correlation is proposed. The constraint equation is established by using the orthogonal relation between the correlation matrix constructed by the received data and the channel vector. Using a small number of known symbols and the improved least square criterion to establish additional constraints, the closed-form solution of the channel response is obtained by the least square method. The algorithm is robust in performance and high in identification accuracy. It is robust to channel noise and channel order error.
【学位授予单位】：解放军信息工程大学
【学位级别】：硕士
【学位授予年份】：2014
【分类号】：TN92
【参考文献】
相关期刊论文 前1条
1 代松银;袁嗣杰;董书攀;;基于子空间分解的信道阶数估计算法[J];电子学报;2010年06期
，本文编号：2276066
本文链接：https://www.wllwen.com/kejilunwen/wltx/2276066.html
