中文摘要：

多输入多输出正交频分复用（multiple input multiple output-orthogonal frequency division muhiple-xing, MIMO-OFDM）系统中，为提升最小二乘（1eastsquares，LS）信道估计算法的估计精度，可以利用离散傅里叶变换（diserete Fourier transform, DFT）算法对LS算法进行改进，然而，DFT算法对最大时延内的噪声未作处理，因此算法的估计精度仍易受噪声影响。针对这一不足，本文提出基于循环前缀（cyclic prefix，CP）的改进DFT信道估计算法。该算法以CP的长度为标准，对CP长度外的信遣系数置零，对CP长度内的信道系数设置门限值，将模值小于门限值的信道系数置零。理论分析和仿真结果表明，改进算法中噪声被有效抑制，估计算法的精度得到提高。

英文摘要：

For the system of multiple input multiple output-orthogonal frequency division multiplexing （MIMO-OFDM）, the channel estimation algorithm using discrete Fourier transform （DFT） can he used to im- prove precision of the least squares （LS） algorithm. However, the DFT based estimation algorithm makes no disposition about the noise existing inside the maximum delay. Therefore, the accuracy of this algorithm is also easily affected by noise. Aiming at this shortage, an improved channel estimation algorithm based on cyclic pre- fix （CP） is proposed. Regarding the length of CP as criterion, the improved algorithm replaces the channel coef- ficients over CP with zero. Meanwhile, it replaces the channel coefficients inside the length of CP with zero when their module values are smaller than the setting threshold value. Theoretical and experiment results prove that the proposed improved algorithm restrains noise effectively and brings some improvement of estimation accuracy.