Estimação via simulações de Monte Carlo em uma classe de sistemas de filas G/G/C

Abstract

Many of the real problems involving queues are characterized by processes in which exact mathematical solutions are not known or are difficult to obtain analytically. In this sense, solutions involving computational analysis, simulations or approximations are of great im- portance. In this research, an algorithm was developed to perform Monte Carlo simulations for general queueing systems and, through the simulated results was obtained estimates for the performance measure of interest of the system. The central idea is to randomly and in- dependently generate two sequences that represent the time between arrivals and the service time. Once the arrival time of each user is determined, as well as his service time, everything that will happen in the system can be known in a deterministic way. To evaluate the results obtained by the algorithm developed in this research, these are compared to some exact or approximate results that are presented throughout the text.