MIMO-OFDM系统中低复杂度重叠的ML信道估计

Overlapped and Low-Complexity ML Channel Estimation for MIMO-OFDM System

构造了一种适于MIMO-OFDM无线系统的低复杂度基于重叠训练图样的ML(Maximum likelihood)信道估计器。通过理论分析:当训练符号矩阵是酋矩阵且训练符号向量摆放在等间距的L子载波上时,重叠ML估计器实现最优的MSE(Mean square error)性能;当训练符号矩阵是归一化Hadamard矩阵时,重叠ML估计器复杂度等于正交。在相关的MIMO移动信道仿真表明:当信道在N个连续OFDM符号时间内保持基本不变时,二者的BER性能是一致的;增加信道时变性后,重叠ML表现了更优的BER(Bit error rate)性能。

A low-complexity ML channel estimator with overlapped training pattern is constructed for MIMO-OFDM wireless system. From theory analysis, this estimator can achieve the optimal MSE performance under the conditions that training symbol vectors are equispaced placement over subcarriers and training symbol matrix （TSM） is a unitary matrix, and its computational amount is equal to that of orthogonal ML estimator when TSM is designed to be a normalized Hadamard matrix. After simulation in mobile MIMO channel, we find the following results： when channel remains approximately invariant during N continuous OFDM symbols, both patterns show the same BER performance; when channel experiences more time varying, the former performs better than the latter.