利用重要性采样的时差-频差联合估计算法

发布时间：2018-05-30 15:23

本文选题：时差 + 频差　；参考：《航空学报》2017年01期

【摘要】：针对无源定位中参考信号真实值未知的时差(TDOA)-频差(FDOA)联合估计问题,构建了一种新的时差-频差最大似然(ML)估计模型,并采用重要性采样(IS)方法求解似然函数极大值,得到时差-频差联合估计。算法通过生成时差-频差样本,并统计样本加权均值得到估计值,克服了传统互模糊函数(CAF)算法只能得到时域和频域采样间隔整数倍估计值的问题,且不存在期望最大化(EM)等迭代算法的初值依赖和收敛问题。推导了时差-频差联合估计的克拉美罗下界(CRLB),并通过仿真实验表明,算法的计算复杂度适中,估计精度优于CAF算法和EM算法,在不同信噪比条件下估计误差接近CRLB。
[Abstract]:In order to solve the problem of TDOA-FDOA-FDOAA joint estimation of unknown true value of reference signal in passive location, a new MLM estimation model is constructed, and the maximum likelihood function is solved by using the importance sampling method. The joint estimation of time difference and frequency difference is obtained. By generating time-frequency-difference samples and calculating the weighted mean of the samples, the algorithm overcomes the problem that the traditional cross-ambiguity function (CAF) algorithm can only obtain integer multiple estimates of sampling intervals in time domain and frequency domain. Moreover, there are no initial value dependence and convergence problems for iterative algorithms such as expectation maximization (EMM) and so on. In this paper, the lower bound of Clemero joint time-frequency difference estimation is derived. The simulation results show that the computational complexity of the algorithm is moderate, the estimation accuracy is superior to that of CAF algorithm and EM algorithm, and the estimation error is close to that of CRLB under different signal-to-noise ratio (SNR).
【作者单位】：解放军信息工程大学导航与空天目标工程学院;
【基金】：国家自然科学基金(61401469,61501513) 国家“863”计划(2012AA7031015)~~
【分类号】：TN911.7

【相似文献】