Ivan Lapatin
Research Subjects
Ivan Lapatin has a total of 14 co-authors. Here are the top co-authors by number of shared publications:
- Anatoly Nazarov
Tomsk State University, Tomsk, Russia
11 shared publications - Svetlana Paul
National Research Tomsk State University
8 shared publications - Olga Lizyura
National Research Tomsk State University
7 shared publications - Alexander Moiseev
Institute of Applied Mathematics and Computer Science, Tomsk State University, Tomsk, Russia
6 shared publications - Bo Bai
Huawei Technologies Company Ltd., Shenzhen, China
4 shared publications - Li Chen
Zhongguancun Laboratory, Beijing, Beijing, China
4 shared publications
Latest Publications
- 25 September 2026
- Published by Novosibirsk State Technical University in Analysis and data processing systems
- No. 3
Characteristics of the output flow of processed tasks in a computational node model under resource competition
The continuous growth in the popularity of remote and cloud computing creates the need for developing mathematical models describing the operation of such systems or their individual components. Mathematical models make it possible to evaluate various performance characteristics of a system under different operating parameters. The calculation of these characteristics is required at the design stage of a computational node or for optimizing the operation of an existing one. The queueing theory is widely used to model processes of a stochastic nature. This paper considers a computational node model represented as a queueing system with an unlimited number of servers. The arrival of tasks to the node is described by a Poisson point process. A distinctive feature of the proposed model compared to the classical one is the dependence of the service (processing) rate of incoming tasks on the total number of tasks present at the computational node. The greater the number of tasks being processed simultaneously, the lower the processing rate of each individual task. This effect arises due to competition for a shared resource of the computational node. In this study, the object of investigation is the output flow of the model, i.e., the moments when task is completed on the computational node. Information about the characteristics of the output flow is of great importance since after processing, tasks (requests) must be transmitted via telecommunication channels. It is the probabilistic characteristics of the serviced task flow that affect the calculation of the required data transmission bandwidth to ensure specified quality-of-service parameters. To analyze the output flow of the computational node model, asymptotic methods of queueing theory are employed. As a result, asymptotic characteristics of the output flow of the proposed model are obtained. It is important to note that the proposed asymptotic approach can potentially be extended to models with non-Poisson task arrival processes.
- 25 September 2026
- Published by Novosibirsk State Technical University in Analysis and data processing systems
- No. 3
Characteristics of the output flow of processed tasks in a computational node model under resource competition
The continuous growth in the popularity of remote and cloud computing creates the need for developing mathematical models describing the operation of such systems or their individual components. Mathematical models make it possible to evaluate various performance characteristics of a system under different operating parameters. The calculation of these characteristics is required at the design stage of a computational node or for optimizing the operation of an existing one. The queueing theory is widely used to model processes of a stochastic nature. This paper considers a computational node model represented as a queueing system with an unlimited number of servers. The arrival of tasks to the node is described by a Poisson point process. A distinctive feature of the proposed model compared to the classical one is the dependence of the service (processing) rate of incoming tasks on the total number of tasks present at the computational node. The greater the number of tasks being processed simultaneously, the lower the processing rate of each individual task. This effect arises due to competition for a shared resource of the computational node. In this study, the object of investigation is the output flow of the model, i.e., the moments when task is completed on the computational node. Information about the characteristics of the output flow is of great importance since after processing, tasks (requests) must be transmitted via telecommunication channels. It is the probabilistic characteristics of the serviced task flow that affect the calculation of the required data transmission bandwidth to ensure specified quality-of-service parameters. To analyze the output flow of the computational node model, asymptotic methods of queueing theory are employed. As a result, asymptotic characteristics of the output flow of the proposed model are obtained. It is important to note that the proposed asymptotic approach can potentially be extended to models with non-Poisson task arrival processes.
- Open Access
- 6 April 2026
- Published by Samara State Technical University
- Vol. 30 (1) , 170-185
Метод асимптотически диффузионного анализа бесконечнолинейной системы массового обслуживания с деградацией скорости обслуживания заявок
Рассматривается математическая модель вычислительного узла в виде бесконечнолинейной системы массового обслуживания с простейшим входящим потоком и интенсивностью обслуживания, зависящей от числа заявок в системе. Указанный эффект называется деградацией скорости обслуживания. Математически деградация описывается положительной невозрастающей функцией, которая определяет зависимость интенсивности обслуживания каждой заявки от суммарного числа заявок, находящихся в системе. Применение данного подхода позволяет более точно моделировать конкуренцию за ресурсы в вычислительных системах, корректно выбирать параметры архитектуры при проектировании новых или оптимизации существующих систем. В то же время зависимость скорости обслуживания от числа заявок существенно усложняет модель и уравнения, описывающие ее функционирование, что требует разработки новых методов исследования. Для анализа предложенной модели применяется метод асимптотически диффузионного анализа в предельном условии высокоинтенсивного входящего потока. Метод реализуется в три этапа и позволяет получить аналитические выражения для основных вероятностных характеристик исследуемой системы. Предложенный метод является приближенным асимптотическим. Однако численные эксперименты и сравнение с допредельными результатами для модели с простейшим входящим потоком демонстрируют малые погрешности и широкую область применимости метода (по значениям параметров модели).
- 6 April 2026
- Published by Samara State Technical University
- Vol. 30 (1) , 170-185
Open AccessМетод асимптотически диффузионного анализа бесконечнолинейной системы массового обслуживания с деградацией скорости обслуживания заявок
Рассматривается математическая модель вычислительного узла в виде бесконечнолинейной системы массового обслуживания с простейшим входящим потоком и интенсивностью обслуживания, зависящей от числа заявок в системе. Указанный эффект называется деградацией скорости обслуживания. Математически деградация описывается положительной невозрастающей функцией, которая определяет зависимость интенсивности обслуживания каждой заявки от суммарного числа заявок, находящихся в системе. Применение данного подхода позволяет более точно моделировать конкуренцию за ресурсы в вычислительных системах, корректно выбирать параметры архитектуры при проектировании новых или оптимизации существующих систем. В то же время зависимость скорости обслуживания от числа заявок существенно усложняет модель и уравнения, описывающие ее функционирование, что требует разработки новых методов исследования. Для анализа предложенной модели применяется метод асимптотически диффузионного анализа в предельном условии высокоинтенсивного входящего потока. Метод реализуется в три этапа и позволяет получить аналитические выражения для основных вероятностных характеристик исследуемой системы. Предложенный метод является приближенным асимптотическим. Однако численные эксперименты и сравнение с допредельными результатами для модели с простейшим входящим потоком демонстрируют малые погрешности и широкую область применимости метода (по значениям параметров модели).
- 24 March 2024
- Published by Springer Nature
- p. 361-372
Approbation of Asymptotic Method for Queue with an Unlimited Number of Servers and State-Dependent Service Rate
- 24 March 2024
- Published by Springer Nature
- p. 361-372
Approbation of Asymptotic Method for Queue with an Unlimited Number of Servers and State-Dependent Service Rate
- 13 May 2023
- Published by Springer Nature
- p. 29-39
Mathematical Modeling of Virtual Machine Life Cycle Using Branching Renewal Process
- 13 May 2023
- Published by Springer Nature
- p. 29-39
Mathematical Modeling of Virtual Machine Life Cycle Using Branching Renewal Process
- 13 May 2023
- Published by Springer Nature
- p. 116-125
Real Waiting Time in Single-Server Resource Queue with Markovian Arrival Process
- 13 May 2023
- Published by Springer Nature
- p. 116-125
Real Waiting Time in Single-Server Resource Queue with Markovian Arrival Process
- Open Access
Queueing System with Two Phases of Service and Service Rate Degradation
In the paper, a queueing system with an unlimited number of servers and two phases of service with degradation in the service rate is studied. The problem of service rate degradation emerges in cloud nodes, where there is contention for hardware resources including computational resources such as CPU cores. In a node, we have a limited number of CPU cores that should execute potentially an unlimited number of processes (requests) in parallel. In our model, the term “server” means a process allocated in the node for execution. So, the number of “servers” is unlimited but their individual performances decrease because CPUs should switch between them during the execution. We consider processes executed in the node with two phases of life cycle that reflects periods with different activity of a process; e.g., in the first phase, the process may require intensive usage of CPU cores but low usage in the second one. Our model distinguishes the phases using different service parameters for them as well as different influence on the service rate degradation in the node. In the paper, two analytical methods are proposed: exact solving of the system of the local balance equation and the asymptotic analysis of the global balance equations. Formulas for the stationary probability distribution of the number of customers in the phases are obtained for both cases. Several numerical examples are provided that illustrate some properties and applicability of the obtained results.
Open AccessQueueing System with Two Phases of Service and Service Rate Degradation
In the paper, a queueing system with an unlimited number of servers and two phases of service with degradation in the service rate is studied. The problem of service rate degradation emerges in cloud nodes, where there is contention for hardware resources including computational resources such as CPU cores. In a node, we have a limited number of CPU cores that should execute potentially an unlimited number of processes (requests) in parallel. In our model, the term “server” means a process allocated in the node for execution. So, the number of “servers” is unlimited but their individual performances decrease because CPUs should switch between them during the execution. We consider processes executed in the node with two phases of life cycle that reflects periods with different activity of a process; e.g., in the first phase, the process may require intensive usage of CPU cores but low usage in the second one. Our model distinguishes the phases using different service parameters for them as well as different influence on the service rate degradation in the node. In the paper, two analytical methods are proposed: exact solving of the system of the local balance equation and the asymptotic analysis of the global balance equations. Formulas for the stationary probability distribution of the number of customers in the phases are obtained for both cases. Several numerical examples are provided that illustrate some properties and applicability of the obtained results.
- 25 February 2022
- Published by Springer Nature
- p. 333-345
Approximation of the Two-Dimensional Output Process of a Retrial Queue with MMPP Input
- 25 February 2022
- Published by Springer Nature
- p. 333-345
Approximation of the Two-Dimensional Output Process of a Retrial Queue with MMPP Input
- 25 February 2022
- Published by Springer Nature
- p. 220-232
Semi-markov Resource Flow as a Bit-Level Model of Traffic
- 25 February 2022
- Published by Springer Nature
- p. 220-232
Semi-markov Resource Flow as a Bit-Level Model of Traffic
- 14 July 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 494-498
Mathematical Model of Scheduler with Semi-Markov Input and Bandwidth Sharing Discipline
In this paper, we consider single server queueing system with multiple semi-Markov inputs and buffers. Each request of the flows brings to the system some random amount of information. According to the bandwidth sharing discipline, each buffer has its own part of the throughput and the server transmits the information from buffers simultaneously. The aim of the current research is to derive the probability distribution of the amount of information in single buffer.
- 14 July 2021
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 494-498
Mathematical Model of Scheduler with Semi-Markov Input and Bandwidth Sharing Discipline
In this paper, we consider single server queueing system with multiple semi-Markov inputs and buffers. Each request of the flows brings to the system some random amount of information. According to the bandwidth sharing discipline, each buffer has its own part of the throughput and the server transmits the information from buffers simultaneously. The aim of the current research is to derive the probability distribution of the amount of information in single buffer.