摘要
在频域通道均衡中,一个关键技术是求解均衡滤波器的响应。文中利用循环卷积和离散傅里叶变换乘积之间的关系,研究了通道均衡中频域除法的两类求解法:一类是时域算法,包括循环矩阵伪逆求解法和循环卷积求循环反卷积算法;另一类是基于最小范数的矩阵形式的频域算法。借助于仿真数据,根据求得的均衡滤波器和待校滤波器的卷积对参考滤波器的逼近误差,验证了上述算法的有效性。实验结果表明,基于最小范数的矩阵形式的频域算法在逼近性能、实时性方面均优于上面两种时域算法。
In frequency-domain channel equalization,one of the most crucial technologies is to seek the response of an equalization filter.According to the relation between the circular convolution in the time domain and the DFT product in the frequency domain,it is referred that two classes of algorithms on frequency-domain division in channel equalization,one class is time-domain algorithms which include an algorithm to seek the pseudo-inverse of a circular matrix and another algorithm to seek the deconvolution by circular convolution,the other is a frequency-domain algorithm having the least MSRe with matrix form.According to the error with the convolution between the filter to be emended and the corresponding equalization filter approximating the reference filter,the validity of the above algorithms is tested.The experimental results shows that the frequency-domain algorithm having the least MSRe with matrix form excels those two time-domain algorithms in both approximation degree and real time.
出处
《现代雷达》
CSCD
北大核心
2011年第4期22-26,共5页
Modern Radar
基金
中国博士后科学基金资助项目(20090461145)
关键词
通道均衡
频域除法
循环反卷积
channel equalization
frequency-domain division
circular deconvolution
