【作者】 龙航；

【导师】 王文博；

【作者基本信息】 北京邮电大学 ， 信号与信息处理， 2010， 博士

【摘要】 无线协同中继技术提供了一种利用空间资源的新手段,多个参与通信的实体之间可通过协作的方式共享有限的资源以提高无线资源的利用率。在无线协同中继系统中,多个终端相互使用对方的天线发送各自的信号,从而形成“虚拟天线阵列”,如此构成的等效多入多出(MMO)系统同样具有较高的系统容量和频谱效率。在传统MIMO系统中,预编码技术是一种应用广泛的链路自适应技术,目前的研究已相对比较成熟。由于协同中继系统与MIMO系统具有相似性,因此可以方便地将预编码技术引入协同中继系统中。本论文研究协同中继系统中的预编码技术,其目标是提升系统的传输可靠性和频谱效率,同时降低信道状态信息(CSI)反馈(共享)开销。本论文中首先总结了无线协同中继通信系统中的预编码技术的研究现状。根据天线阵或虚拟天线阵的构成方式,我们可以将无线协同中继通信中的预编码技术分为3类,即单中继节点空域预编码技术、多节点分布式空域预编码技术和时域预编码技术。每一类预编码技术可以进一步根据系统拓扑结构细化为更多子类。本论文中对3类无线协同中继系统中的预编码技术都进行了深入研究：论文的第二部分关注第1个类别：协同中继系统中的单中继节点预编码技术。在线性转发的单中继系统中,我们首先关注了简单的迫零(ZF)中继系统,提出在中继节点的前向和反向ZF滤波器之间增加一个空间信道配对或映射矩阵,用以降低转发噪声功率。空间信道配对矩阵适用于单个源节点时,空间信道映射矩阵适用于单个目标节点时。由于空间信道配对或映射的使用提升了每个数据流的接收信干噪比,因此系统的可靠性和有效性可以同时得以提升。其次,我们将现有线性转发的单源单中继单目标系统概括为一个统一框架。基于此统一框架,我们提出了完整的中继节点的空间信道映射矩阵设计方法：在中继节点的前向和反向滤波器之间增加一个空间信道映射矩阵,用以降低中继转发噪声功率或目标节点的处理后噪声功率。多种单中继系统下的理论分析和仿真结果证明了统一框架下的空间信道映射对系统性能的提升。论文的第三部分关注第2个类别：协同中继系统中的多节点分布式预编码技术。根据源节点的发送信号形式可以将此类别分为广播模式和单播模式。首先,我们研究了各种广播模式的典型方案,找出了其中系统容量性能较为卓越的基于QR分解(QRD)的广播方案。我们提出了两种改进基于QRD的广播方案的方法：行列置换优化和分布式预编码(中继节点功率分配)。前者的思路是对信道矩阵进行行列置换,QRD的结果是不一样的。我们提出了一种改进的矩阵QRD算法,在分解的同时选择矩阵的列置换方案,以极小的计算复杂度代价获得了系统容量性能的明显提升。后者是将分布式预编码引入基于QRD的广播方案中,在所有的中继节点发送功率一定时,最优的分布式预编码矢量搜索问题等价为一个多维空间中的球面上的最优化问题。为此,我们提出了一种新的粒子集群优化(PSO)算法用于多维球面上的最优化搜索问题。基于提出的球面上的PSO算法,基于QRD的广播方案中的分布式预编码(中继节点功率分配)可以有效的提升系统容量,且计算复杂度在可接受的范围内。其次,我们关注了多中继系统中的单播模式,提出和分析了多种单播模式预编码方案,对比了单播模式和广播模式各自的应用场景和性能优劣,给出了典型单播方案下的接收信噪比(SNR)的解析表达式,提出了基于单播模式的中继节点选择策略,简化了CSI反馈,降低了系统开销,取得了系统容量性能和反馈开销间的折中。论文的第四部分关注第3个类别：协同中继系统中的时域预编码技术。在协同中继系统中,利用源节点在广播阶段和多址接入阶段的发送符号(矢量)作为虚拟天线阵,可以引入时域预编码技术。在单天线协同中继系统中,为了最大化系统容量,我们提出了时域预编码与多节点分布式空域预编码的联合优化。在多天线协同中继系统中,由于时域预编码技术的使用需要系统中所有链路的CSI,我们提出了基于ZF中继协议的CSI压缩和反馈机制,将每个中继节点相关的等效信道信息和转发噪声信息压缩为两个正实数,极大的降低了CSI反馈开销。根据对归一化等效信道矩阵的奇异值的分析,我们给出了时域预编码技术可以支持的数据流数量的上界,并分析了影响时域预编码性能的各种因素的优先级。更多还原

【Abstract】 The wireless cooperative/relay technique introduces a new type of spatial diversity in communication systems. In wirelss cooperative/relay communication systems, terminals share system resource with others to enhance the system reliability and capacity. The so-called "virtual antenna array" is constructed, so that the virtual multi-input-multi-output (MIMO) system model can be formulated, which can achieve high spectral efficiency. As one of the promoting link adaptation techniques, precoding attracts much attention recently in MIMO systems and has been widely investigated. Due to the similarity between cooperative systems and MIMO systems, precoding techniques can also be used in cooperative systems. The subject of this dissertation is the investigation of precoding techniques in wireless cooperative/relay systems. Precoding designs for the system reliability and the spectral efficiency are considered. The implementation complexity and the channel state information (CSI) requirement are also considered in the research.The existing precoding techniques in wireless cooperative/relay systems are firstly reviewed and discussed in this dissertation. According to the construction of antenna arrays, precoding in cooperative/relay systems can be summarized as three categories:1. the local spatial precoding with single relay node; 2. the distributed spatial precoding through multiple nodes; 3. the temporal precoding through successive transmitted signals at the source node. Each category can be further divided into several sub-categories according to the implementation scenarios. The three categories are all investigated in this dissertation.The second part of this dissertation deals with the local spatial precoding. Firstly, in the linear relaying systems with single relay node, we focus on the zero-forcing (ZF) relaying system. A spatial channel pairing or mapping matrix is proposed between the backward ZF filter and the forward ZF filter to reduce the relaying noise. The spatial channel pairing matrix is suitable for systems with single source node, while the spatial channel mapping matrix is suitable for systems with single destination node. Since the spatial channel pairing or mapping matrix improves the received signal-to-interference-and-noise ratio of each data stream, the system reliability and the spectral efficiency are both enhanced. Secondly, a unified framework is presented, including all kinds of linear relaying systems with single-source single-relay single-destination. Based on the unified framework, the complete design of spatial channel mapping matrix is presented, i.e. the spatial channel mapping matrix is presented at the relay node between the backward filter and the forward filter to reduce the relaying noise or the local noise at the destination node. Analysis and simulation results in several systems with single relay node demonstrate the advantage of the spatial channel mapping matrix design.The distributed spatial precoding through multiple nodes is investigated in the third part of this dissertation. According to the transmitting signal of the source node, there are two modes of precoding in multi-relay systems, i.e. the broadcasting mode and the unicasting mode. Firstly, precoding schemes in the broadcasting mode are studied, and the QR decomposition (QRD) based scheme is chosen since it performs well on the system capacity. Two methods for improving the QRD based scheme are presented:1) permutation optimization; 2) distributed spatial precding (power allocation among relay nodes). The first method utilizes the different decomposition results of matrix QRD with column permutation. A modified QRD algorithm is presented to decompose a matrix with column permutation optimization. The permutation optimization with the modified QRD algorithm obviously improves the system capacity with slight complexity increase. The second method utilizes distributed precoding through multiple relay nodes. With the assumption that the powers of all relay nodes are constrained as a whole, the distributed precoder design is equivalent to an optimization problem on a spherical surface. A novel particle swarm optimization (PSO) algorithm on the spherical surface is presented to enhance the system capacity with acceptable complexity. Additionally, the unicasting mode is studied. Several unicasting schemes are proposed and analyzed. Application scenarios and capacity performance of the broadcasting and the unicasting modes are compared. With a typical unicasting scheme, the exact expression of the received signal-to-noise ratio (SNR) is derived. A relay node selection strategy based on the unicasting mode is presented, where the CSI feedback is simplified.The last part of this dissertation discusses the temporal precoding in cooperative/relay systems. Temporal successive transmitted symbols (vectors) at the source node in the broadcasting phase and multi-accessing phase are treated as a virtual antenna array. In multi-relay systems with single-antenna nodes, the temporal precoding at the source node and the distributed spatial precoding through relay nodes are jointly designed to maximize the system capacity. In cooperative/relay systems with multi-antenna nodes, the CSI of all links in the system are required at the source node. A CSI compression and feedback scheme basing on the ZF relaying protocol is presented to support the temporal precoding at the source node, where the equivalent channel and the relaying noise information related to each relay node are compressed into two positive real coefficients. According to the analysis of the singular values of the normalized equivalent channel matrix, the upper bound of the number of data streams supported by the temporal precoding is derived. Various factors affecting the performance of the temporal precoding are also analyzed.更多还原