基于多层核空间的BPSK信号频率估计方法

Frequency Estimation Method of BPSK Signal Based on Multiplayer Kernel Space

在强噪声背景下,针对BPSK调制信号,提出了一种新的频率估计技术。该估计器的设计是基于多层分类核主成分分析及改进核主成分分析,并利用不同的特征空间距离测量法。基于多层核主成分提取估计器需要将调制信号的训练样本根据各自的频率进行分层。为了获得精确的频率估计,估计器首先根据分层结构,提取样本信号的特征从而来获得被观察信号的初始频率。基于该初始估计,来建立最佳的分层处理结构,并根据训练样本的特征空间的最优选择实现特征提取函数的改进。在仿真结果中作了上述两个算法的比较来验证理论的应用。同时该结果也显示在低信噪比下的估计器的卓越性能。

A new frequency estimation technique for BPSK modulation signals embedded in strong noise is proposed.The design of the proposed estimator is based on multilayer kernel classification principal compo-nent analysis（MLKPCA）or improved kernel component principal analysis（KPCA）together with distance in feature space measure methods.The estimator based on kernel principal component extraction requires to stratify the training samples of interested signals with respect to their respective frequencies.In order to achieve accurate frequency estimation,the estimator first extractes the feature of the sample signals to get an initial frequency corresponding to the stratification structure.Based on the initial estimation,an optimal mul-tilayer classification processing structure is then constructed.Comparison of the two algorithms through the simulation results are also made to corroborate the theoretical application.The results also show that signifi-cant performances the estimator in extreme low signal-to-noise ratio can be obtained.