Sergei Shepelenko

Faculty: computer information technologies and automation (CITA)

Department: automation and telecommunication (AT)

Speciality: telecommunication systems and networks

Theme of master's work:

Development and research traffic control systems in MPLS networks

Scientific advisor: Ph.D., assistant professor of the department AT Vladimir Bessarab

Development and research of traffic control systems in the MPLS networks

Currency

The developing of the telecommunications is the most rapid area in the world nowadays. New IP - services such as IPTV, VoIP and others are developing like in telecommunications networks. In connection with this IP - traffic has increased significantly. Therefore, penetration of CoS (class of service) for different traffic types is a necessary task. It is also necessary to introduce a system of traffic management that will provide specified classes of service for different types of traffic by the network resource redistribution. Use of multiprotocol label switching MPLS (MultiProtocol Label Switching) technology on transport level makes it possible to ensure the efficient traffic transmission to support the parameters of QoS (Quality of Service). Traffic management possibilities which concentrate at ensuring of the balanced using of MPLS network resources with QoS support are implemented by the Traffic Engineering (TE) technology with help of balanced loading mechanism of network resources, selecting the optimal route of traffic, usage of backup procedures and distribution the network loading, balancing traffic and mechanisms of overloading prevention.

Communication of work with scientific programs, plans, themes.

Master's qualification thesis is executed during 2010-2011 agrees with a scientific direction of the department “Automation and Telecommunications” of Donetsk National Technical University.

The purpose of master thesis

The purpose of master’s thesis is to improve MPLS network efficiency with help of effective resource allocation of throughput between a set of specified routes in backbone channels and with help of load redistribution between the LSP in conditions of changing network traffic.

Tasks

It is necessary to solve several tasks to achieve this goal:

- to analyze the MPLS network management system;

- to develop the traffic control system in MPLS with use of neural network models;

- to develop the predictive control system;

- to investigate QoS requirement, which we are lodged to the MPLS tunnels.

The main part

MPLS network architecture, which based on the LSR (Label Switching Router), provides a fast packet switching by the usage of labels in the packet header. The labels store the delivery address and network level class (Forwarding Equivalence Class, FEC). Using unidirectional TE-tunnels is the mechanism for specifying the path of traffic passing on MPLS-TE tunnel. TE tunnel combines a sequence of LSR, that’s selected with a glance the maximum network resources loading and the QoS requirements. In order to provide fault-tolerant routing in the MPLS-TE network we use a technology called fast reroute packets Fast ReRoute (FRR), which allows you to send traffic temporarily to a spare pre-configured TE-tunnel. Spare tunnel is configured to reroute the traffic capacity in the case of route fail. The choice of TE-tunnel is defined with specified criteria, such as the minimum packet delay [1,2].

The important advantage of MPLS is an opportunity to transfer not the packet with a single label but a stack of labels within the MPLS architecture.

Addition / removal operation of tags are defined as operations on the stack (push / pop).

Only the upper label stack sets result of switching , the bottom stack is passed transparently before the upper label stack will be strike off. This approach allows us to create a flows hierarchy in the MPLS network and organize the transfer tunnel.

Thus, the main feature of MPLS is the separation packet switching process from the IP-addresses analysis in the header, which opens up a number of catching opportunities [5].

Using neural networks

In the near future I’m going to create a method for estimating the load on the line by using neural networks.

Network is trained to give some set of inputs for set of outputs. Each input (or output) set is regarded as a vector. Training is carried out by successive presentation of input vectors with simultaneous adjustment of weights in accordance with certain procedures. In the process of network training the weights has gradually become such that each input vector makes the output vector.

The set is formed from the data load. We can see the picture below:

Conclusion

Using neural network models for forecasting traffic in the communication channel will allow us to use the channel resources more efficiently when we control the traffic. This will lead to satisfy the QoS requirements. Thus we can ensure a specified quality of service, which operator provide to users.

Literature

1. П.П. Воробієнко, Л.А. Нікітюк, П.І. Резніченко Телекоммунікаційні та інформаційні мережі: Підручник для вищих навчальних закладів –К: САММІТ-КНИГА, 2010. – 640 с.
2. Оліфер В.Г. Оліфер Н.А. Компьютерные сети. Принципы, технологии, протоколы: Учебник для вузов 3-е изд. – СПб: Питер, 2008 – 958 стр
3. Гольдштейн А. Б Проблемы перехода к мультисервисным сетям
4. . Сподах К.М. Глобальные сети [Электронный ресурс] - Режим доступа к статье: http://revolution.allbest.ru/radio/00307902_0.html
5. Гольдштейн А. Б. Механизм эффективного туннелирования в сети MPLS. Вестник связи. —№2 — 2004.
6. Семенов Ю.А. Telecommunication technologies - телекоммуникационные технологии [Электронный ресурс] Ю.А. Семенов/ - Режим доступа к статье: http://book.itep.ru
7. Базовые сервисы технологии MPLS [Электронный ресурс] - Режим доступа к статье: http://www.rublin.org.ua/article/bazovye-servisy-tekhnologii-mpls
8. И.В. Баскаков, А.В. Пролетарский, Р.А. Федотов, С.А. Мельников IP-телефония в компьютерных сетях ИНТУИТ.РУ, Бином.Лаборатория знаний, Интернет-Университет Информационных Технологий (2008) – 112с.
9. Мишенин А.А. Основы искусственных нейронных сетей [Электронный ресурс] Мишенин А.А. / - Режим доступа к статье: http://neuronets.chat.ru/foundations.html
10. Комашинский В.И., Смирнов Д.А. – Нейронные сети и их применение в системах управления и связи, 2002;

Note

When the author was writing this abstract the master’s qualification thesis wasn’t been completed. Date of final completion of work: December, 15, 2011. Full text and materials of the work can be received from the author or his scientific supervisor after that date. Abstract in Russian gives more expand description of investigated problems.