Abstract: TDOA estimation is very important in case of multi-station passive location，but the traditional co-correlation method is difficult to be used to track fast moving target because of non real-time，high computation cost and lower accuracy.This paper proposes a new method based on Homotopy Continuation Extended Kalman Filter（HCEKF），which be used to quickly converge to CRLB in performance.In the new approach，the TDOA parameter is updated along with a new observation，and the phenomena of non-convergence in general EKF is eliminated when initial state estimation be not closed to true parameter.

Key words: TDOA estimation, co-correlation, signal model, HCEKF

摘要： 在无源时差定位中，要求能实时获得目标到达测量站的TDOA参数，传统的互相关法需要大数据量、大运算量以及数据块处理，难以提取快速时变参数。提出了一种基于同伦延拓EKF的新方法，收敛速度快，估计精度达到CRLB下限，对每一个新的数据样点可直接更新TDOA参数，克服了一般EKF法中初始估计误差过大将导致滤波发散的问题。

关键词: 时差估计, 互相关, 信号模型, 同伦延拓EKF