论文部分内容阅读
本文介绍了加快重要取样收敛的被改进的重要采样(imPortance Sa-mpling)技术。利用这种可以估算雷达模拟中低虚警概率的方法,蒙特卡洛运算次数明显地减少了。对于一维指数、书布尔和瑞利分布,可获得一个均匀无偏最小方差估算器。对于高斯分布,用这种方法得到的估算器同样比先前已知重要采样方法的好。而对于单元平均系统,用这种技术与分组取样相结合的方法,在20个参考单元和10~(-6)虚警率的情况下,与先前的已知重要采样方法比较,蒙特卡洛运算次数可减少170次。
This article describes imPortance Sampling technology that speeds up convergence of important samples. Using this method, which estimates the probability of low false alarm in radar simulations, the number of Monte Carlo operations is significantly reduced. For a one-dimensional index, Book Boolean and Rayleigh distributions, a uniform and unbiased minimum variance estimator can be obtained. For Gaussian distributions, the estimator obtained in this way is also better than previously known important sampling methods. For the average cell system, this technique combined with packet sampling method, 20 reference cells and 10 ~ (-6) false alarm rate, compared with the previously known important sampling method, Monte Carlo The number of operations can be reduced by 170 times.