空间干扰对齐的信号检测与性能分析研究
发布时间：2018-07-17 07:03
【摘要】：无线通信的主要目标之一是提高系统的频谱效率,而干扰则是实现该目标的主要限制因素之一,它不但影响频谱效率的提高,而且缺乏有效的处理方案。最近,关于干扰处理技术的研究取得了较大进展,即干扰对齐(Interference Alignment,IA)。研究表明干扰不是无线通信的根本限制。理论上,IA在干扰信道中,可以获得K/2倍的复用增益,其中K为干扰信道中的用户数,即IA的信道容量可以与用户数成线性关系,这使得IA引起了广泛的关注。IA通过预编码技术,使得在每个目的节点,所有的干扰相互重叠,从而为期望信号提供无干扰的信号维度,使彻底消除干扰对期望信号的影响成为可能。本学位论文的主要研究内容包括四个方面：IA的闭式预编码算法、迭代预编码算法、分集算法以及基于随机矩阵理论的IA性能分析。内容安排如下：首先,给出了实现干扰对齐的条件,并证明了干扰对齐解的非唯一性。在某些情况下,干扰对齐的预编码矩阵可以直接获得,但由于干扰对齐的目标函数并不是信道容量,且IA问题的多解性,使得IA解的优化比较困难。针对这个问题,分析了IA中期望信号与干扰子空间的内积范数的特点,得到了二者的内积范数与处理后期望信号功率的关系。提出了内积范数最小的干扰对齐算法,分析表明该算法可以优化期望信号能量,从而提高系统的信道容量。注意到IA的闭式预编码矩阵只能在特定场景下才能获得,一般情况下,IA的预编码矩阵只能通过迭代方法获得。IA的迭代算法首先建立一个目标函数,通过迭代最小化该目标函数来得到预编码矩阵。典型的迭代干扰对齐算法是分布式IA,而最大信干噪比(Max-Signal-to-Interference-Noise-Ratio, Max-SINR)算法可以在一定程度上作为干扰信道的容量限。Max-SINR算法的主要问题是不能保证收敛,而且复杂度也很高。通过研究Max-SINR算法不收敛的原因,同时又借鉴了该算法的思想,分析干扰子空间及期望信号子空间的各自特点及相互关系,提出以干扰与期望信号泄漏出相应子空间功率的加权和作为目标函数,通过迭代最小化该目标函数来得到IA的预编码矩阵,而且证明了该算法是收敛的。进而,分别就权值为常数以及权值可调情况,讨论了各自算法的性能。结果表明,虽然与Max-SINR算法相比,加权算法的性能,在低SNR时仍有差距,但较分布式IA,性能有明显提高,而且权值可调算法的性能在低SNR时要明显优于固定权值算法。由于IA的关注点在于最大程度地消除干扰对期望信号的影响,因而其在低信噪比时的性能较差。针对这个问题,本学位论文研究了分集IA的实现方法及性能。首先给出了分集IA的实现条件,然后以接收分集为例研究了单侧分集的实现方法并通过仿真给出了其性能。接着,探索了同时在收发两侧同时实现分集的可能性,并提出了实现方案,给出了初步的分析结果。结果表明,在同样的天线配置条件,分集可以有效地提高IA在低信噪比时的性能,而在高信噪比时,复用的性能要优于分集。为了得到IA在复杂通信系统中的性能,迫切需要得到它的理论近似。而随机矩阵为物理层提供了良好的理论近似,也适用于干扰信道的研究。本文基于随机矩阵理论,给出了IA信道容量的理论预测。由于IA的核心思想在于消除干扰对期望信号的影响,根据这个特征,首先用渐近特征分析的方法,得到IA等效矩阵特征值的分布特点,给出了IA理论上的系统性能。然后,又得到了IA系统容量的确定性等价矩阵,并用不动点方程的方法成功对该问题求解。数值仿真表明,以上两者都得到正确的预测结果,且后者的适用范围更广,能够用于分集干扰对齐的研究,且不要求矩阵的极限存在。
[Abstract]:One of the main objectives of wireless communication is to improve the spectral efficiency of the system, and interference is one of the main limiting factors to achieve this goal. It not only affects the enhancement of spectral efficiency, but also lacks effective processing schemes. Recently, the research on interference processing technology has made great progress, that is, interference alignment (Interference Alignment, IA). Research shows that interference is not the fundamental limitation of wireless communication. In theory, IA can gain K/2 times multiplex gain in the interference channel, in which K is the number of users in the interference channel, that is, the channel capacity of IA can be linear with the number of users, which causes IA to attract extensive attention to.IA through precoding technology, so that in each destination node, The main research contents of this dissertation include four aspects: the closed precoding algorithm of IA, the iterative precoding algorithm, the subset algorithm and the IA performance based on the random matrix theory. The contents are as follows: first, the condition of interference alignment is given, and the non uniqueness of the interference alignment solution is proved. In some cases, the precoding matrix of the interference alignment can be obtained directly, but because the target function of the interference alignment is not the channel capacity, and the multi solution of the IA problem makes the optimization of the IA solution more difficult. In view of this problem, the characteristics of the inner product norm of the expected signal and the interference subspace in IA are analyzed. The relationship between the inner product norm of the two and the signal power in the later stage is obtained. The interference alignment algorithm with the minimum inner product norm is proposed. The analysis shows that the algorithm can optimize the expected signal energy and improve the capacity of the system. The closed precoding matrix to IA can only be obtained in a specific scene. In general, the precoding matrix of IA can only obtain an iterative algorithm of.IA by iterative method. First, a target function is set up, and the precoding matrix is obtained by iterative minimization of the target function. The typical iterative interference alignment algorithm is distributed IA, and the maximum is the largest. The main problem of the Max-Signal-to-Interference-Noise-Ratio (Max-SINR) algorithm, which can be used as the capacity limit of the interference channel to a certain extent, is that it can not guarantee the convergence and the complexity is very high. By studying the reason of the non convergence of the Max-SINR algorithm, and using the idea of the algorithm, the interference subspace is analyzed. The respective characteristics and relations of the subspace of the signal and the expected signal are presented. The weighted sum of the corresponding subspace power of the interference and the expected signal is given as the target function, and the precoding matrix of the IA is obtained by minimizing the objective function iteratively, and it is proved that the algorithm is convergent. Then, the weights are constant and the weights are available, respectively. The performance of each algorithm is discussed. The results show that, although compared with the Max-SINR algorithm, the performance of the weighted algorithm still has a gap at low SNR, but the performance of the distributed IA is obviously improved, and the performance of the weight adjustable algorithm is obviously superior to the fixed weight algorithm when the SNR is low. Because the focus of IA is to eliminate the maximum degree. Interfering with the effect of the expected signal, so its performance is poor at low signal to noise ratio. For this problem, this dissertation studies the implementation and performance of diversity IA. First, we give the realization conditions of diversity IA. Then, taking the reception diversity as an example, we study the implementation of the single side diversity and give its performance through simulation. At the same time, the possibility of realizing diversity at the same time at both sides of the receiver is presented, and the implementation scheme is proposed and the preliminary analysis results are given. The results show that the diversity of the IA can effectively improve the performance of the same antenna configuration when the signal to noise ratio is low, while the performance of the multiplexing is better than the diversity at high signal to noise ratio. In order to get IA in the complex communication system, The performance of the system is urgent to obtain its theoretical approximation. The random matrix provides a good theoretical approximation to the physical layer and is also suitable for the study of the interference channel. Based on the random matrix theory, the theoretical prediction of the capacity of the IA channel is given. Because the core idea of IA is to eliminate the influence of interference to the desired signal, according to this feature First, the characteristic value distribution of the IA equivalent matrix is obtained by the asymptotic characteristic analysis method, and the system performance in the IA theory is given. Then, the deterministic equivalent matrix of the capacity of the IA system is obtained, and the solution of the problem is solved by the fixed point equation. The numerical simulation shows that the above two get the correct prediction results. The latter is more suitable for the study of diversity interference alignment and does not require the existence of matrix limit.
【学位授予单位】：电子科技大学
【学位级别】：博士
【学位授予年份】：2014
【分类号】：TN911.4
，
本文编号：2129520
[Abstract]:One of the main objectives of wireless communication is to improve the spectral efficiency of the system, and interference is one of the main limiting factors to achieve this goal. It not only affects the enhancement of spectral efficiency, but also lacks effective processing schemes. Recently, the research on interference processing technology has made great progress, that is, interference alignment (Interference Alignment, IA). Research shows that interference is not the fundamental limitation of wireless communication. In theory, IA can gain K/2 times multiplex gain in the interference channel, in which K is the number of users in the interference channel, that is, the channel capacity of IA can be linear with the number of users, which causes IA to attract extensive attention to.IA through precoding technology, so that in each destination node, The main research contents of this dissertation include four aspects: the closed precoding algorithm of IA, the iterative precoding algorithm, the subset algorithm and the IA performance based on the random matrix theory. The contents are as follows: first, the condition of interference alignment is given, and the non uniqueness of the interference alignment solution is proved. In some cases, the precoding matrix of the interference alignment can be obtained directly, but because the target function of the interference alignment is not the channel capacity, and the multi solution of the IA problem makes the optimization of the IA solution more difficult. In view of this problem, the characteristics of the inner product norm of the expected signal and the interference subspace in IA are analyzed. The relationship between the inner product norm of the two and the signal power in the later stage is obtained. The interference alignment algorithm with the minimum inner product norm is proposed. The analysis shows that the algorithm can optimize the expected signal energy and improve the capacity of the system. The closed precoding matrix to IA can only be obtained in a specific scene. In general, the precoding matrix of IA can only obtain an iterative algorithm of.IA by iterative method. First, a target function is set up, and the precoding matrix is obtained by iterative minimization of the target function. The typical iterative interference alignment algorithm is distributed IA, and the maximum is the largest. The main problem of the Max-Signal-to-Interference-Noise-Ratio (Max-SINR) algorithm, which can be used as the capacity limit of the interference channel to a certain extent, is that it can not guarantee the convergence and the complexity is very high. By studying the reason of the non convergence of the Max-SINR algorithm, and using the idea of the algorithm, the interference subspace is analyzed. The respective characteristics and relations of the subspace of the signal and the expected signal are presented. The weighted sum of the corresponding subspace power of the interference and the expected signal is given as the target function, and the precoding matrix of the IA is obtained by minimizing the objective function iteratively, and it is proved that the algorithm is convergent. Then, the weights are constant and the weights are available, respectively. The performance of each algorithm is discussed. The results show that, although compared with the Max-SINR algorithm, the performance of the weighted algorithm still has a gap at low SNR, but the performance of the distributed IA is obviously improved, and the performance of the weight adjustable algorithm is obviously superior to the fixed weight algorithm when the SNR is low. Because the focus of IA is to eliminate the maximum degree. Interfering with the effect of the expected signal, so its performance is poor at low signal to noise ratio. For this problem, this dissertation studies the implementation and performance of diversity IA. First, we give the realization conditions of diversity IA. Then, taking the reception diversity as an example, we study the implementation of the single side diversity and give its performance through simulation. At the same time, the possibility of realizing diversity at the same time at both sides of the receiver is presented, and the implementation scheme is proposed and the preliminary analysis results are given. The results show that the diversity of the IA can effectively improve the performance of the same antenna configuration when the signal to noise ratio is low, while the performance of the multiplexing is better than the diversity at high signal to noise ratio. In order to get IA in the complex communication system, The performance of the system is urgent to obtain its theoretical approximation. The random matrix provides a good theoretical approximation to the physical layer and is also suitable for the study of the interference channel. Based on the random matrix theory, the theoretical prediction of the capacity of the IA channel is given. Because the core idea of IA is to eliminate the influence of interference to the desired signal, according to this feature First, the characteristic value distribution of the IA equivalent matrix is obtained by the asymptotic characteristic analysis method, and the system performance in the IA theory is given. Then, the deterministic equivalent matrix of the capacity of the IA system is obtained, and the solution of the problem is solved by the fixed point equation. The numerical simulation shows that the above two get the correct prediction results. The latter is more suitable for the study of diversity interference alignment and does not require the existence of matrix limit.
【学位授予单位】：电子科技大学
【学位级别】：博士
【学位授予年份】：2014
【分类号】：TN911.4
，
本文编号：2129520
本文链接：https://www.wllwen.com/kejilunwen/wltx/2129520.html
