In this paper, we review a retrial queue with MMPP input and two-way communication. Incoming calls, arriving at the server and finding it busy, join orbit and try to enter the server again after some exponentially distributed time. While idle, the server makes outgoing calls and serves them for exponentially distributed time with another intensity. MMPP (Markov Modulated Poisson Process) is an input process, in which control is driven by a continuous Markov chain. Changing its state entails a change of the intensity of the arrival process. For this model we present an asymptotic approximation of the two-dimensional characteristic function under the condition of a large delay of requests in the orbit. For this approximation we carried out a numerical experiment, where asymptotic results were compared to computations, which were obtained via simulation.
Распределенные компьютерные и телекоммуникационные сети: управление, вычисление, связь DCCN-2021 [Электронный ресурс] : материалы XXIV Международной научной конференции (20-24 сентября 2021 г., Москва, Россия). М., 2021. С. 369-375