MAXIMUM CONNECTION COUNT ESTIMATION DEVICE, METHOD AND PROGRAM

摘要

PROBLEM TO BE SOLVED: To estimate a parameter for the distribution of service times from an average value per observation cycle on the basis of periodic observations on the number of applications connected, and calculate an α probability point (0α1) of the distribution of maximum connection counts or an approximate value thereof.;SOLUTION: An arrival ratio and a service time distribution parameter are estimated from plural average connection counts obtained by periodic observations. Then, on the basis of the identification result, an M/G/∞ model is constructed which is composed of M/G/∞ parameters for turning the number of connected applications into a model, a cumulative distribution of maximum connection counts within any finite time in the M/G/∞ model is calculated, and the α (0α1) probability point thereof is estimated. On the basis of the identification result, the M/G/∞ model is constructed, a cumulative distribution of a rate between the maximum connection count and the average connection count in each periodic observation is calculated, and the α (0α1) probability point is estimated. Further, the adaptability of observed values is tested against the estimated service time distributions, and, out of a plurality of parametric service time distributions which have not been rejected, a distribution deemed to be most suitable is selected.;COPYRIGHT: (C)2013,JPO&INPIT