摘要
通过联合优化训练符号设计和ML准则,构造了一种基于重叠训练图样的拥有低PAPR的时频信道估计器ML-A+LI.通过理论分析获得:当训练符号矩阵等于归一化Hadamard矩阵且在子载波上等间距的摆放时,ML-A+LI估计器实现等同于正交的ML信道估计器ML-B+LI最优的MSE性能和低复杂度.在M IMO移动信道仿真表明:ML-A+LI和ML-B+LI的BER性能介于LMMSE+LMMSE和LI+LI之间,更接近于LMMSE+LMMSE;当信道变化较慢时,ML-A+LI和ML-B+LI具有相同的BER性能,当信道变化较快时,ML-A+LI比ML-B+LI提供了更优的BER性能.
A low-complexity time-frequency channel estimator maximum likelihood （ML）-A ＋ LI （linear interpolation） with overlapped training pattern （TP） and low peak-to-average power ratio （PAPR） is constructed by virtue of both optimal training pattern design and ML rule. It can achieve the optimal mean square error （MSE） performance and low complexity as ML-B ＋ LI using orthogonal TP under the conditions that training symbol vectors are equispaced placement over subcarfiers and training symbol matrix is a normalized Hadamard matrix. From simulation and analysis in multi- ple input-multiple output （MIMO） mobile channel, the following facts are found： ① the bit error ration （BER） performance of ML-A ＋ LI and ML-B ＋ LI is worse than that of LMMSE ＋ LMMSE （linear minimum mean square error） and better than that of LI ＋ Li, but is more close to that of LMMSE ＋ LMMSE;
② ML-A ＋ LI achieves the same performance as ML-B ＋ LI for slow car speed whereas the former performs better than the latter for fast car speed.
出处
《东南大学学报（自然科学版）》
EI
CAS
CSCD
北大核心
2006年第5期700-704,共5页
Journal of Southeast University：Natural Science Edition
基金
国家自然科学重大基金资助项目(60496311)
东南大学移动通信国家重点实验室开放课题资助项目(200609)
关键词
多输入多输出技术
正交频分复用
重叠
最大似然
信道估计
multiple input-multiple output
orthogonal frequency division multiplexing
overlapped
maximum likelihood
channel estimator
