第1次自己回帰モデルによるゆらぎの解析

概要

論文の詳細を見る
1/f noises, the power spectrum density (PSD) of signals inversely proportional to the frequency, have been observed in various types of physical and physiological systems; e.g. current and heart beat. 1/f noise is suggested to be generated by superimposing stochastic processes. Among stochastic processes, autoregressive process is suitable for analyzing physical and physiological systems in that the autoregressive process is time-discrete. A time-discrete model is suitable for analyzing those systems because those systems are observed in time-discrete manner as well. However, it is obscure whether 1/f noises are generated by superimposing those autoregressive processes. In this study, first order autoregressive (AR(1)) processes were superimposed with varied the process parameter and a driving noise. As a result, PSD of superimposed time sequences was inversely proportion to the frequency when the process parameter and a driving noise were a uniformly distribution and white noises, respectively. This result suggests that 1/f noises could be generated from the superimposing AR(1) processes when the process parameter is distributed uniformly.