基于LS-GPR的OFDM信道估计算法  

Channel estimation algorithm of OFDM system based on LS-GPR

作　　者：吴明[1] 宋博[1] 王锋[1] 唐红[1] 

机构地区：[1]空军工程大学航空航天工程学院,陕西西安710038

出　　处：《计算机工程与设计》2016年第1期1-4,共4页Computer Engineering and Design

基　　金：国家自然科学基金项目(61379104)

摘　　要：针对传统快时变信道下的OFDM信道估计性能低下问题,提出基于梳状导频的最小二乘高斯过程回归(LS-GPR)时变信道频率估计算法。介绍高斯过程回归理论,它是一种具有强大的非线性拟合能力和小样本学习能力的机器学习方法。通过LS估计出的导频处的频率响应用作训练序列,使用高斯过程回归方法估计其它子信道的频率响应。仿真结果表明,在多径信道情况下,LS-GPR信道估计算法明显优于传统的信道估计方法,在高信噪比时误码性能有很大提升。Aiming at the poor estimation performance of the traditional orthogonal frequency division multiplexing in fast time-varying channel,an OFDM channel estimation algorithm based on least square Gaussian process for regression（LS-GPR）using comb-type pilot was proposed.The theory of Gaussian process for regression was introduced.The frequency response at pilot channel estimated through LS method was used as training data and the frequency response at other sub-channels were estimated using GPR method.The simulation results show that under the multipath channel circumstance,the LS-GPR channel estimation algorithm shows obviously better performance compared to traditional ones and it promotes BER performance greatly under high SNR.

关 键 词：正交频分复用 高斯过程回归 信道估计 最小二乘 插值 

分 类 号：TN911.5[电子电信—通信与信息系统]