Parameter estimation of K distribution based on second-kind statistics
基于第二类统计量的K分布参数估计

In order to efficiently estimate the parameters of K distribution, this paper proposed the log-cumulant estimator. Based on second-kind statistics, first it derived the second-kind first characteristic function of K distribution by taking the Mellin transformation to the probability density function of K distribution, second obtained the second-kind second cha-racteristic function of K distribution from the logarithmic transformation of the second-kind first characteristic function, and last deduced the first two log-cumulants to estimate the parameters of K distribution by taking the derivative of the second-kind second characteristic function. Compared to the traditional maximum likelihood estimator, the log-cumulant estimator of K distribution with analytical expressions was easy to compute. Monte Carlo simulations demonstrate that the log-cumulant estimator of K distribution based on second-kind statistics achieves high estimation accuracy.