Моделі та методи оцінки ефективності управління наданням інтелектуальних сервісів в мережах наступного покоління

Abstract

UA: У дисертаційній роботі досліджуються питання оцінки ефективності управління наданням інтелектуальних сервісів (ІС) в мережах наступного покоління. Удосконалено метод формування комплексного критерію ефективності управління наданням ІС (ЕУНІС), який, на відміну від відомих, дозволяє врахувати особливості управління наданням ІС, а також надає можливість кількісної оцінки ЕУНІС. Набув подальшого розвитку метод оцінки функціональної живучості інтелектуальної надбудови NGN, що дозволяє визначити ймовірність виконання хоча би однієї функції деякого класу ІС. Уперше розроблено аналітичну модель інтелектуальнї надбудови з централізованим принципом управління наданням ІС з урахуванням самоподібності потоку заявок на ІС, що надає можливість визначити потрібні мережні ресурси для забезпечення необхідного значення ЕУНІС. Набули подальшого розвитку методи побудови імітаційних моделей процесів управління наданням ІС інтелектуальною надбудовою, які, на відміну від існуючих, дозволяють урахувати самоподібність трафіку на інтелектуальні сервіси.

EN: The thesis examines the questions of an estimation of management efficiency for providing intelligent services (IS) in next-generation networks. As a result of telecommunications analysis, the role and place of IS in the process of transition from the intelligent network to the concept of NGN and the concept of IMS is determined. It is proposed to evaluate the management effectiveness of the provision of IS (EMPIS) by a complex criterion, which includes the following sub-criteria: the time of service, the probability of refusal to provide the service, the length of the queue, which receives the application for service, structural and functional survivability and structural reliability, as well as cost intelligent superstructure. The method of formation of the complex criterion of EUPIS is improved, which, unlike known ones, allows to take into account the features of IP management, and also makes it possible to quantify the effectiveness of the principles of managing the provision of intelligent services with an intelligent superstructure. The method of estimating the functional survivability of the intelligent addition NGN has been further developed, which makes it possible to determine the probability of performing at least one function of a certain class of intelligent services. The nature of the flow of applications for IS is investigated. For research, the R/S method is implemented, with the help of AutoSignal software. The analysis was carried out on the network of the company, which provides hosting services sites. The results of traffic measurements were obtained using the Zabbix software. It has been found that traffic that contains IS applications can range in time - from several hours to a year. The analytical model is an intelligent superstructure with a centralized management principle of providing IS with considering the self-similarity of the flow of request for IS for the first time, which provides the ability to identify the necessary network resources that are needed to provide the necessary value of the management effectiveness of providing intelligent services. The method of calculating the sub-criteria of the EMPIS criterion is given. The analytical model of intelligent superstructure presented in the form of queuing system with a homogeneous flow of applications is proposed. In order to take into account the self-similarity of traffic, a function has been introduced that depends on the Hurst coefficient. The analytical model of the intelligent superstructure presented in the form of the queuing system with K classes of applications without priorities is also proposed. The criterias EMPIS of intelligent superstructure are calculated. The methods for constructing simulation models for managing the provision of an intellectual superstructure by IS have been further developed, which, unlike existing ones, allows one to take into account the self-similarity of traffic to intelligent services. For simulation models of intelligent superstructure with exponential traffic, the general purpose simulation system (GPSS) (General Purpose Simulation System) was used. NS2 (network simulator) is used to develop simulation intelligent superstructure models taking into account the self-similarity of traffic. The simulation models of intelligent superstructure with one and three classes of applications are proposed without taking into account and taking into account the self-similarity of traffic. It is determined that the value of the criterion EMPIS provision taking into account the self-similarity of traffic decreases in comparison with the exponential nature of traffic of at least 38%, that is, with the self-similar nature of traffic, the effectiveness of management of IS provision significantly deteriorates. Thus, in the design of NGN and its subsystem - intelligent superstructure, the mandatory condition is the study of the nature of predicted traffic in the network and its consideration in determining the required network resources.