基于稀疏表示的大规模MIMO系统信道估计和传输方案设计的研究

The channel estimation is a fundamental problem for the signal detection and adaptive transmission, which largely determines the performance of communication systems. The major problems confronted by the channel estimation for massive MIMO systems are: 1) It is very difficult to estimate the downlink channel as the base station (BS) has a large amount of antennas. 2) Pilot contamination, caused by the increasing number of mobile stations (MS), will severely restrict the channel estimation performance. As demonstrated by the current works, the sparse representation channel estimation method is expected to solve these two problems simultaneously. To cope with the downlink channel estimation and pilot contamination, in this project, we aim to fully discuss the sparse characteristics of channel state information matrix for massive MIMO systems, and try to develop several sparse representation methods for the channel estimation, especially, by using sparse Bayesian learning and low-rank matrix representation. In order to further improve the transmission performance of massive MIMO systems, we also plan to give some analyses of the relationship among the channel estimation, the transmission scheme and the system performance in detail, and then propose some sparse presentation methods for joint optimization of the channel estimation and transmission scheme. Through the studies in the project, we are expected to obtain some original scientific research achievements so as to promote the development of channel estimation and transmission optimization for massive MIMO systems.

信道估计是通信信号检测和自适应传输的基础，对通信系统的性能起着至关重要的作用。在大规模MIMO系统中，信道估计面临的主要问题是：1）基站端天线数较多，下行链路的信道估计异常困难；2）随着用户数的增加，导频污染将严重制约信道估计的性能。现有研究表明基于稀疏表示的信道估计算法有望能同时解决上述两大问题。因此，本项目拟深入分析大规模MIMO系统信道具有的稀疏特性，研究和完善基于稀疏表示方法（特别是稀疏贝叶斯学习和低秩矩阵表示方法）的信道估计算法，以应对大规模MIMO系统信道估计中下行链路信道估计和导频污染两大挑战。同时探讨信道估计、传输方案、系统性能之间的相互关系，在保持合理计算复杂度的情况下，联合优化基于稀疏表示方法的信道估计和传输方案设计，进一步改善大规模MIMO系统的传输性能。通过本项目的研究，有望取得原创性的科研成果，在一定程度上丰富和发展大规模MIMO系统信道估计和传输方案的优化设计。

经过四年的研究与探索，本项目较全面地分析了大规模MIMO系统信道具有的稀疏特性，研究和完善了基于稀疏表示方法的信道估计算法，同时深入探讨了信道估计、传输方案、系统性能之间的相互关系，并获得了一些信道估计和系统联合优化设计方案，进一步改善了大规模MIMO系统的传输性能。项目主要取得的成果有：1）大规模MIMO通信系统信道的稀疏特性研究方面取得重大突破，提出了三种的具有创新性和实用性的稀疏信道模型；2）提出了若干基于稀疏贝叶斯学习方法的大规模MIMO系统信道估计方法；3）提出了一种基于低秩矩阵表示方法的大规模MIMO系统信道估计方法；4）提出了若干具有鲁棒性的大规模MIMO系统信道估计方法；5）揭示了效优先原则下多用户MIMO系统的能量分配特征；6）提出了若干高精度的波达方向估计方法。本项目超额完成了既定的研究任务，在项目执行期间，申请国家发明专利14项，其中授权6项；正式发表SCI论文20篇，其中信号处理领域TOP期刊4篇。通过本项目培养了青年教师2名、博士研究生2名和硕士研究生6名。