Abstract: In Orthogonal Frequency Division Multiplexing（OFDM） systems, in order to solve the problem that the common channel estimation algorithms cannot suppress the noise in the channel impulse response within the length of cyclic prefix, an improved channel estimation algorithm based on Discrete Fourier Transform（DFT） is proposed, which works in an iterative fashion. The initial estimation of the channel is first calculated by the Least Square（LS） channel estimation algorithm based on the pilot sub-carriers. Then after IDFT, the distribution of channel impulse response in time domain can be obtained through the energy growth rate function which can be used to suppress the noise. Through the iterations, inter-carrier interference and additive white Gauss noise can be effectively depressed. Simulated with the LS channel estimation algorithm, the traditional channel estimation algorithm based on DFT and the channel estimation algorithm based on DFT with the threshold, the estimation performance of the proposed algorithm is best no matter the Doppler shift is small or large. When the bite error rate is [10-2], the system performance of the improved channel estimation algorithm based on DFT improves 3~5 dB compared with other algorithms.

Key words: Orthogonal Frequency Division Multiplexing（OFDM）, channel estimation, Discrete Fourier Transform（DFT）, iterative estimation, Doppler shift

摘要： 在正交频分复用（OFDM）系统中，针对常用的信道估计算法不能有效地抑制信道冲激响应中循环前缀长度内噪声的不足，提出了一种改进的基于离散傅里叶变换（DFT）的信道估计算法。该算法是一个多次迭代的过程，通过最小二乘算法获得导频位置处的信道频域响应，经过逆傅里叶变换后，利用时域内引入的能量增长速率函数来判断信道冲激响应分布情况，以便对其进行消噪处理，最后通过多次迭代进一步抑制子载波间干扰和加性高斯白噪声。仿真结果表明，无论在多普勒频移较小还是较大的情况下，该算法的估计性能均优于最小二乘（LS）信道估计算法、传统基于DFT的信道估计算法和基于阈值的信道估计算法。在系统误比特率为[10-2]时，改进的基于DFT的信道估计算法比其他算法有3~5 dB的性能增益。

关键词: 正交频分复用（OFDM）, 信道估计, 离散傅里叶变换（DFT）, 迭代估计, 多普勒频移