基于FFT计算α平稳过程概率密度的改进算法

Improved FFT-based alpha-stable density approximation algorithm

在网络流量建模和参数估计的过程中,α平稳过程的概率密度计算是一项重要的基础工作。结合α平稳过程特征函数的数学特性,在以往基于FFT求解α平稳过程概率密度函数算法的基础上做出如下改进:通过计算自动选择合适的采样区间和采样间隔;引入扩频进一步降低计算复杂度。实验表明,改进算法比传统算法计算复杂度更低,并可以有效控制计算误差。

Approximation of density of α-stable process is an elementary works in modeling network flow using α-stable process. Combining the mathematical properties of characteristic function of α-stable process with the traditional approximating algorithm, an improved FFT-based algorithm was proposed. The improved algorithm has 2 noticeable features automatically choosing the sampling space and sampling interval, importing frequency expansion. Experiments show that it can decrease calculation complexity and effectively control computation error.