时延估计的高分辨方法

High resolution methods for time delay estimation

目前，时延估计方法的研究热点集中在自适应解卷积方面，但解卷积方法的分辨率下限是数据的采样间隔，在短数据情况下无法获得足够的分辨率。本文首先指出了常用时延估计方法的局限性，然后提出了两种时延估计的高分辨方法：时延频率估计和特征结构法。这两种方法的最小分辨率都可以突破数据采样间隔。本文详细论述了它们的估计原理及性能特点，并给出了计算机仿真结果。最后，列表比较了多种时延估计方法的性能．

Recently the most researches in time delay estimation are concentrated on adaptive deconvolution.It's well known that minimum delay resolution in deconvolution methods is the sampling interval of original data. It is impossible to obtain enough delay resolution in the case of short data records. In this paper, the limitation of commonly used delay estimation methods are pointed out firstly. Then two delay estimation methods with high resolution are presented, which named delay--frequency estimation and eigenstructure method. It's indicated that these two methods can resolve delays spaced closer than sampling interval. The principle and characteristics of them are described in detail, and the corresponding simulated resultS are given. At the last, a comparison table is listed to show the performance of main methods in time delay estima tion.