互相关法时延估计中的多径抑制方法

浅海信道是一个多径信道，对于基于相关法的延迟估计，多径效应常常会导致相关函数中出现伪峰，伪峰的幅度有时会超过主峰，进而增加了相关峰选取的难度。利用 AR 法滤波对信号中的多径分量加以抑制，从而削弱了伪峰的幅度，提高了延迟估计的准确性。先从理论上论证了线性预测法可以对多径干扰进行抑制，再用仿真信号和实际信号验证了方法的效果。对于仿真信号，在无噪声干扰的条件下，选择合适阶数的滤波器可以使信号中的多径分量被完全抑制；对于正信噪比的实际信号，其中的多径分量在一定程度上被抑制，从而由多径分量产生的相关峰会被明显削弱，减少了峰值位置的跳变。

The multipath effect eliminating technique for time delay estimation of cross-correlation method

Shallow water channel is a multipath channel. For time delay estimation of cross-correlation method, multipath effects usually lead to secondary peaks, of which the magnitude may be larger than the main peak of the correlation function. And, this increases the difficulty of choosing the right peak. In this paper, AR filter is used for eliminating the multipath components and softening the secondary peaks to increase the accuracy of time delay estimation. Theoretical analysis proves that the linear prediction method can eliminate the multipath effect, and then the effect of this method is testified with both simulated and real signals. For simulated signals, the linear prediction method with appropriate orders can completely eliminate the multipath components of the original signals. For real signals with positive SNR, the mul-tipath components can be eliminated to a certain extent, thus the secondary peaks caused by multipath components will be significantly softened and main peak shifting will be reduced.