基于最小二乘平滑算法的时变信道盲辨识

Blind identification of time-varying channels using least squares smoothing

针对基展开模型的时变信道盲辨识问题,提出了联合阶数估计的最小二乘平滑算法。该算法充分利用其投影误差矩阵包含的时变信道阶数和系数信息,在信道辨识前先进行阶数精确估计,克服了该子空间类盲辨识方法对信道阶数敏感的缺点,改善了辨识效果;与线性预测类方法相比,该方法不需要知道输入信号的统计特性。理论推导和仿真结果验证了该算法的有效性。

According to the blind identification problem of the basis expansion model of time-varying channel, a joint order detec- tion and blind channel estimation by least squares smoothing algorithm is proposed. The projection error matrix contained the in- formation of the order and coefficient is exploited to estimate the channel order before channel identification in this algorithm, the shortcoming that the subspace blind identification method are sensitive to the channel order is overcomed and the performance of the blind identification is improved; this method does not need to know the statistical features of the input signal compared with linear prediction methods. Theory and simulation results prove the effectiveness of the algorithm.