基于贝叶斯滤波及平滑理论的动态协作中继单跳信道估计研究

The cooperative relay networks can utilize the individual channel state information to optimize the relaying operation and thus enhance the system performance. Most existing works assume the cooperative relay system experiences the static channel fading. However, due to the motion of the source, relay or destination, the cooperative relay network is more susceptible to time-varying fading as compared to the conventional point-to-point communication systems, which makes the existing static individual channel estimation methods no longer valid. In this project, we will embrace the time correlations of the dynamic individual channels, and utilize the basis expansion model to construct initial individual channel acquisition algorithm, where the topology structures of the cooperative relay networks are fully taken into consideration. We will then apply Bayesian filtering theory to construct proper filters to track the individual channel dynamic variations online. Moreover, we will resort to Bayesian smoothing theory to design corresponding smoothers for the individual channel tracking results in an offline manner. Hence, the dynamic individual channel estimation is implemented at destination without the aid of channel information feedback links. Furthermore, we will analyze the individual channel estimation errors during the initial acquisition, the tracking and the smoothing, based on which we will design robust training sequences and calculate the optimal training power allocation. The proposed project targets at developing the dynamic individual channel tracking-smoothing system for the relay network, which can serve as the theoretical foundation and technical support for the future cooperative relay networks.

在协作中继网络各节点处获取单跳链路的信道信息可优化中继信号处理操作，从而提升系统全局性能。目前绝大多数协作中继信道估计研究均假设信道服从准静态衰落，然而信源、中继和信宿间易发生相对移动，致使信道发生动态变化，导致现有的准静态单跳信道估计不再适用。本项目将结合动态单跳信道的时间相关特性，基于有限参数基扩展模型设计切合协作中继系统自身拓扑架构的动态单跳信道信息初始捕获算法，进而融合中继节点的辅助功能，运用贝叶斯滤波和平滑理论实现在线信道跟踪和离线信道平滑等。所设计的算法无需节点间信道信息反馈链路，可以有效的节省频谱资源。在此基础上，分析动态单跳信道初始捕获、跟踪及平滑误差的统计特性，设计鲁棒性的训练序列，寻求最优训练功率分配方案，最终形成动态协作中继单跳信道的完整估计体系，为未来协作中继系统应用提供重要的理论依据和技术支撑。

项目对协作中继系统信宿端的动态单跳信道估计技术进行了深入和系统的研究。取得的研究成果如下：.1) 针对多天线中继系统在中继叠加导频框架提出一种迭代式线性最小均方误差独立信道估计算法。从理论上证明了所提出的迭代算法的收敛性。所提算法收敛速度快，估计精度高，而且计算复杂度很低。推导出了两跳独立信道的最小二乘估计闭式解，作为迭代算法的初始值。此外，推导了估计误差的两个下界：贝叶斯克拉美罗界和线性估计下界。基于LELB设计了最有导频结构和最优中继放大转发因子，以提高估计精度。.2) 研究处于时间选择性平坦衰落场景中的放大转发单向中继网络的信道估计问题，其中网络中各中继节点采用分布式空时编码技术。首先，为降低网络中未知信道参数向量的维数，采用多项式基扩展模型将单跳信道估计转换为单跳信道扩展模型系数向量的估计。其次，借助自回归模型，构建单跳信道扩展模型系数向量的动态状态空间。然后，采用无迹卡尔曼滤波器实现单跳信道扩展模型系数向量的前向动态跟踪，相应的利用无迹Rauch-Tung-Stribel平滑器后向平滑前向跟踪的结果。最后，推出单跳信道扩展模型系数向量估计向量的贝叶斯克拉美罗界。.3) 研究处于时间选择性平坦衰落场景中的放大转发单向中继网络的信道估计贝叶斯克拉美罗界，其中，中继节点采用叠加导频训练框架以便于信宿获取单跳信道信息。构建单跳信道的非线性动态状态空间，推导全数据辅助信道估计器和部分数据辅助信道估计器的在线或离线贝叶斯克拉美罗界。特别的，对于部分数据辅助信道估计器，假设信源数据映射为方形QAM集合，设计数值量化评估在线或离线贝叶斯克拉美罗界的框架，分析信源采取4-QAM调制时贝叶斯克拉美罗界在高信噪比下的渐进性能。