为了解决低信噪比长伪码直扩信号的盲估计难题,提出了一种长伪码直扩信号的相关矩阵累加平均结合特征分析的新方法。本方法是在已知长伪码直扩信号的扩频码周期、码速率等参数,并且信息序列不相关的前提下,将接收到的长伪码直扩信号以一随机确定值为起点进行周期分段形成连续多个观察向量,求其相关矩阵并累加平均,并实施特征分析以得到相关矩阵的特征值和信号所含主成分,由相关矩阵的特征值可以估计出信号所含的噪声方差、信号信噪比和信号分段时随机确定的起点,由主成分特征向量还可以进一步估计出观察信号的扩频码序列。本文所提出的方法充分分析和推导了长伪码直扩信号的特征结构,提出了对长伪码直扩信号盲估计的核心基础算法,该算法可以有进一步应用于长伪码直扩信号盲解扩的潜力。本文提出的方法并不象已经提出的其他算法,其对长伪码扩频信号的盲估计性能可以随着观察向量个数的增加逐步得到改善,而且该方法可以应用于任意类型的扩频码、信息码,在理论上能工作在任意强度的加性高斯白噪声环境下,并且不需要事先提取任何定时同步信息。理论分析和数值结果表明了本方法较为鲁棒不易受到噪声影响。
In this paper,we propose an approach of correlation matrix accumulation and eigen-analysis for long code DS signals, which can estimate the lower signal to noise ratios (SNR) long code DS signals blindly. Of course, the parameters of the long code DS signals (such as period and chip interval of the PN sequence) need to be known. We divide the received signals into vectors according to the period of PN sequence,and the start point of division is a randomized certain point uniformly distributed in a period of PN sequence. Then we calculate and accumulate the correlation matrix of signal vectors, and compute its eigen-analysis. We can estimate the variance of noise in signals, signal to noise ratios and the start point of division from the eigenvalues of the correlation matrix, furthermore we can estimate the PN sequence of long code DS signals from the main component eigenvector blindly. Based on the estimated PN sequence, we can realize the long code DS signals despreading without the PN sequence. The theoretic analysis and experimental results show that the approach is very robust,it can work well on lower SNR input signals under common circumstances.