Channel estimation is always implemented in communication systems to overcome the effect of interference and noise. Especially, in wireless communications, this task is more challenging to improve system performance while saving resources. This paper focuses on investigating the impact of geometries of antenna arrays on the performance of structured channel estimation in massive MIMO-OFDM systems. We use Cram'er Rao Bound to analyze errors in two methods, i.e., training-based and semi-blind-based channel estimations. The simulation results show that the latter gets significantly better performance than the former. Besides, the system with Uniform Cylindrical Array outperforms the traditional Uniform Linear Array one in both estimation methods.

翻译：信道估计始终在通信系统中用于克服干扰和噪声的影响。尤其在无线通信中，这一任务在提升系统性能的同时节约资源更具挑战性。本文重点研究天线阵列几何结构对大规模MIMO-OFDM系统中结构化信道估计性能的影响。我们利用克拉美-罗界分析两种方法（即基于训练的信道估计与基于半盲的信道估计）的误差。仿真结果表明，后者显著优于前者。此外，采用均匀圆柱阵的系统在两种估计方法中均优于传统的均匀线阵。