FDD制式大规模MIMO系统信道信息获取问题和理论方法研究

As an important technical support of 5G system, the potential performance benefits of massive MIMO highly depend on the acquisition of the downlink channel state information (CSI) at the base station (BS). For the frequency division duplexing (FDD) MIMO system, the pilot overhead for the CSI acquisition is prohibitive due to the channel non-reciprocity, and the transmission characteristics determines that the resource at the BS will be shared between the CSI acquisition stage and the data transmission stage. Apart from these, there will be non-ideal conditions, e.g., the hardware impairments, which all make the CSI acquisition face many challenges. This project investigates the CSI acquisition problem of FDD massive MIMO system in terms of the pilot overhead, the resource utilization, and the robustness requirements. Firstly, starting from the original number of BS antennas, the asymptotic performance of transmission rate is analyzed with number of antennas and the number of pilots changing at different speeds by using the random matrix theory such that we can obtain the principles of setting pilot overhead based on antenna selection. Then, the non-orthogonal pilot signal with low overhead is optimized, and the closed-loop estimation strategy is designed to improve the estimation accuracy. Secondly, the energy efficiency is introduced, and the joint resource allocation of channel estimation stage and data transmission stage is conducted with the aid of optimization theory and game theory. Finally, the non-ideal factors are modeled, and the impacts of them on the CSI acquisition are analyzed. According to the different robustness requirements, we realize the low cost, energy efficient and robust scheme for CSI acquisition.

大规模MIMO作为5G系统的重要技术支撑，其性能潜力有赖于基站对下行信道信息的获取。FDD制式大规模MIMO系统因上下行信道非互易，致使信道信息获取时导频开销过大，且其传输特点又决定了信道信息获取与数据传输要共享基站传输资源，加之硬件损伤等非理想条件，使得信道信息获取问题面临诸多挑战。本项目拟从导频开销、资源利用率、鲁棒性三方面对FDD大规模MIMO系统信道信息获取问题进行研究。首先，从基站天线数本源出发，利用随机矩阵理论分析天线数与导频数以不同速率变化时系统速率渐进性能，获得基于天线选择的导频开销设置准则，进而优化低开销非正交导频信号，并采用闭环联合估计策略提高估计精度。其次，引入能效指标，借助最优化理论和博弈论等，对信道估计和数据传输资源联合分配。最后，对非理想性因素进行建模，综合分析其对信道信息获取造成的影响，根据不同的鲁棒性需求，优化实现低开销、高能效和强鲁棒的信道信息获取方案。

大规模MIMO技术作为5G/B5G的重要关键技术，具有提升系统容量、简化波束设计、降低发射功率等诸多优点，而上述优势需要基站获取准确的下行信道状态信息。FDD制式下信道互易性条件的缺失使得大规模MIMO下行信道估计的导频开销骤增，而下行传输过程决定了基站还需要在信道估计与数据传输两个环节进行合理的资源分配，以保证信道估计精度和传输有效性。同时，在高速移动等非理想条件下，信道呈现严重的时变性，会影响信道信息获取精度，并最终体现在系统性能上。因此，本项目围绕大规模MIMO系统，在FDD制式下导频开销分析和导频信号设计、能效资源分配、非理想时变信道下的系统性能分析等方面展开研究。具体内容包括：(1)基于天线选择的导频开销分析和导频信号设计。分析了下行导频开销随天线数以不同速率变化时对信道估计精度和系统传输速率的影响，基于此获得了天线数增长时导频开销的设计准则。进一步，设计了以最大化系统下行速率为目标，松弛列正交约束的新型导频信号，并得到了最优导频信号的闭合形式解；(2)能效目标下的信道估计与数据传输联合资源分配优化。以最大化系统能效为目标，以导频时长与功率以及数据功率等系统资源为参量，并考虑系统功率负载和最小频谱效率约束，综合评估信道估计精度、系统频谱效率以及功率消耗性能；(3)非理想时变信道条件下的系统频谱效率性能分析。以一阶高斯马尔科夫过程刻画信道时变性特性，推导系统平均频谱效率关于信道时间相关系数的解析表达式，分析时间相关系数对系统频谱效率渐近极限性能和发射功率缩放增益所产生的影响。本项目的研究成果有助于解决大规模MIMO系统信道信息获取时存在的若干关键问题，并为下一代无线通信系统的设计提供了一定的理论基础和方法支撑。