论文部分内容阅读
最大似然类的无源定位方法可以达到较高的定位精度,但其计算量非常大。由于时频参数联合定位模型本身的非线性和非凸性非常大,繁重的计算量在TOA与FOA联合定位系统中表现尤为明显。本文针对这一问题,通过Pincus全局最优理论和蒙特卡洛重要性采样技术降低了最大似然类定位算法的计算复杂度,并且保证算法可以收敛到全局最优解。本文主要的贡献是构建了一个高斯分布的概率密度函数来近似原始的代价函数方便后续的采样,我们称之为重要性函数。该方法所带来性能上的提升是因为选择了最优的重要性函数并且Pincus保证算法收敛到全局最小值。这一处理大大降低了计算量,由于算法进行了泰勒级数展开,需要初始估计值。通过采样处理并且对样本进行加权,本文算法对初始估计值具有良好的鲁棒性。最后,实验证明本文所提算法可以达到克拉美罗限,且性能要优于现有算法。
The maximum likelihood class of passive location method can achieve higher positioning accuracy, but its calculation is very large. Due to the very large non-linear and non-convexity of the time-frequency joint location model itself, the heavy computation load is particularly obvious in TOA and FOA joint positioning system. In order to solve this problem, this paper reduces the computational complexity of the maximum likelihood class locating algorithm by Pincus global optimization theory and Monte Carlo importance sampling algorithm, and guarantees that the algorithm converges to the global optimal solution. The main contribution of this paper is to construct a Gaussian distribution probability density function to approximate the original cost function to facilitate subsequent sampling, which we call the importance function. The performance improvement of this method is due to the selection of the optimal importance function and Pincus guarantee that the algorithm converges to the global minimum. This process greatly reduces the amount of computation, which requires initial estimates due to the Taylor series expansion of the algorithm. Through the sampling process and the sample weighting, the proposed algorithm has good robustness to the initial estimate. Finally, experiments show that the proposed algorithm can achieve the Keramide limit, and the performance is better than the existing algorithms.