Abstract: Aiming at the poor performance of Frequency Domain Space-Alternating Generalized Expectation-Maximization（FD-SAGE） algorithm in channel estimation and the slow convergence speed in the MIMO-OFDM system, this paper presents an improved FD-SAGE channel estimate algorithm. On the basis of FD-SAGE algorithm, a modified SAGE algorithm is derived through decomposing and analyzing the potential and incomplete data of SAGE algorithm. Meanwhile, the maximum likelihood algorithm is introduced in the process of refreshing data information in SAGE algorithm, and then the reliability of system is improved. The theoretical analysis and simulation results show that this algorithm can better track the channel change and acquire a faster convergence speed, just in cost of a slight system complexity. The performance is better than the traditional Least Square（LS） algorithm. Comparing with the ideal channel estimation under the maximum likelihood algorithm in signal detection, the new proposed algorithm has only a loss of 0.5 dB with the same bit error rate.

Key words: MIMO-OFDM system, SAGE algorithm, improved FD-SAGE algorithm, channel estimation

摘要： 针对MIMO-OFDM系统中频域的空间交替广义期望最大化（FD-SAGE）算法估计信道性能较差以及收敛速度慢的问题，提出了一种改进的FD-SAGE信道估计算法。该算法在FD-SAGE算法的基础上，通过对SAGE算法的潜在数据和不完全数据进行分解分析推导出一种修正的SAGE算法，同时在SAGE的更新数据信息时引入最大似然算法，进而提高系统的可靠性。理论研究和仿真结果表明，该算法以牺牲少量复杂度为代价，能较好地追踪信道变化且收敛速度较快，其性能优于传统的LS算法，信号检测采用最大似然算法时，在相同误比特率情况下与理想信道估计仅相差0.5 dB。

关键词: MIMO-OFDM系统, SAGE算法, 改进的FD-SAGE算法, 信道估计