CC BY
21
MODEL OF THE PROCESS OF THE MANAGEMENT SYSTEM AND THE CARRIER ETHERNET NETWORK ELEMENTS INTERACTION
Elina V. Login,
St.Petersburg state transport university of the emperor Alexander I, St.Petersburg, Russia, elinabeneta@yandex.ru
Andrey K. Kanaev,
St.Petersburg state transport university of the emperor Alexander I, St.Petersburg, Russia, kanaevak@mail.ru
DOI 10.24411/2072-8735 -2018-10010
Keywords: сarrier Ethernet, agents, multi-agent systems, multi-agent systems, sequential evaluation algorithm, MathCAD environment.
Carrier Ethernet (CE) is a technology for building a new generation network. This technology is characterized by high scalability and complexity of the processes performed. Such processes include a set of mechanisms implemented within the CE technology for controlling and managing both the state of network elements and its configuration. The mechanisms for controlling and managing the state of CE network elements (OAM mechanisms) do not currently have a formalized description for the management system (MS). To solve this problem, it is necessary to develop a multiagent MS model with the Carrier Ethernet network. This, in turn, is impossible without the presence of a complex of models demonstrating a number of processes in the interaction of network elements and components of a multi-agent MS. The developed model of the process of interaction of the control system with elements of the transport network CE is obtained using the sequential estimation algorithm, and the simulation result is demonstrated with the help of the MathCAD computing environment. The formation of multiple types of messages for the transmission of control data is the heart of the key stage of modeling. It is based on functional model of multi-agent MS of transport network CE. The initial data for calculating the objective function of the model contains a set of sub processes. They included algorithms for controlling and managing the state of the transport network CE elements. These algorithms were developed earlier. The results of the study are presented in the form of dependencies of the control cycle time from various parameters. In this case the control cycle time is determined in the process of controlling and managing the state of the network elements, and such parameters will be considered control information transmission parameters and MS structural parameters. The reasonably chosen values of the considered parameters allow us to formulate requirements for individual processes and the entire MS as a whole. The results of the model presented in the article can be the basic data for simulation modeling of separate control and management processes for the elements of the CE transport network, as well as the basis for the development of a methodology for building a multi-agent MS for the transport network of CE.
Information about authors:
Elina V. Login, St.Petersburg state transport university of the emperor Alexander I, Assistant of the department "Electrical Communication", St.Petersburg, Russia
Andrey K. Kanaev, St.Petersburg state transport university of the emperor Alexander I, Head of the department "Electrical Communication", St.Petersburg, Russia
Для цитирования:
Логин Э.В., Канаев А.К. Модель процесса взаимодействия системы управления с элементами сети Сarrier Еthernet // T-Comm: Телекоммуникации и транспорт. 2018. Том 12. №1. С. 47-52.
For citation:
Login E.V., Kanaev A.K. (2018). AModel of the process of the management system and the Сarrier Еthernet network elements interaction. T-Comm, vol. 12, no.1, pр. 47-52.
í i л
Introduction
With the increase in the scale of the networks and, as a result, the increased structural and functional difficulties, the development in the direction of the corresponding management systems (MS) is being actively pursued. In this regard, a new technology for building trans port-based networks has emerged, namely Carrier Ethernet, a carrier-class network that is highly complex in its processes, and must be implemented in large-scale networks. To dale, there is no single methodological basis for building a network management system for a new generation of Carrier Ethernet. The proposed multi-level structure of modem MS by the communication network proposed [1| includes three main levels of management: organizational, operational-technical and technological levels. In this paper, it is proposed to use the agent management apparatus at the operational and technical level, which will allow the promising MS to control and manage the states of the network elements of a large scale. This multiagent management system (MAMS) for interaction with elements of the Carrier Ethernet communication network has a set of processes for controling and managing the states of network elements [2, 3]. The paper provides a model for preliminary estimation of the duration of the interaction process between the MAMS and the communication network.
Interaction scheme between MAMS and Carrier Ethernet
The functional model of the MAMS by the CE transport network is created in [1]. So the interaction between the MAMS and CE is performed by means of three manage blocks that are part of each MAMS node. These blocks perform the functions of data collection and interaction with own CE elements (ARE(J\ as well as data collection and interaction between each oilier (Astate and Asm)- These data are further used to form a fragment of the CE structure when calculating the objective function (OF) of the multi-agent community.
In the framework of this paper, it is actual to estimate the control cycle time for the state of network elements. For this it is convenient to use sequential estimation algorithms. The RobbinsMonroe algorithm is the most suitable of these kinds of algorithms from the point of view of computational complexity, accuracy and speed of the results obtained [4,5].
Forming multiple types of messages
for the transmission of control data
During the implementation of the control circuit is possible to distinguish the following types of transmitted data in the interaction of management agents and CE elements:
- interrogation by the agent of registration and analysis events block (Areoí) of the element parameter values and the subsequent response of the element. This amount of data will be denoted by Vp and the frequency of this operation - The amount of data after interrogation the parameter is composed of a request and a reply, as shown in expression (1),
V =Vn" + Vrep (1)
p p p y '
- the agent ARFG, to do requests for execution of control commands - LBM and LTM (requests for CE commands for checking the bi-directional connectivity of the route and localization of the fault in the route), as well as the subsequent transfer of data on the application of these commands to the CÉ fragment - LBR and LTR (response messages to the requested the CE comands). Designate these commands as OAM, since
they are implemented by OAM Carrier Ethernet mechanisms. The entire amount of data transmitted during the execution of control commands will be determined by the expression (2).
V =Vreq +Vrcr' (2)
r oam r oam oam vti!
- notification (without a preliminary interrogation) by the CE element of the Arec^ agent. It is possible also add alarm indication notifications from the CE fragment element - AIS. This amount of data will be denoted by V„ and the frequency of this operation - v".
It is important to note that the management agents interact with other control blocks of the MAMS, thereby organizing internal interaction. Let is say that the only possible type of interaction between control blocks and agents is the types of "interrogation", "data" and "notification" messages. Then, the types of messages that must be transferred between the control blocks are created:
- interrogating (on demand) by the state evaluation block {Astate) of the agent of registration and analysis events block (Areo) about the state of some element that is in the subordination of this network fragment. In this case, the amount and frequency of the transmitted message will be denoted by Vp and v , respectively. The amount of data in an agent's interrogation is composed of a request and a reply, as shown in expression (3).
V' =V'ri'" +V'rep (3)
p p p
- data received by the Astate block on notifications of the CE element, in particular various kinds of failures and information on the verification of connectivity of routes (for example, CCM termination);
- notification (without a preliminary interrogation) of the structure formation block (AStr) about the appearance of structural changes in the CE fragment. Denote this amount of data and the frequency of this operation by the variables Vs and V1.
Formalized representation of the objective function
Supposed that the process of forming a solution variant is started inside the Asm block. Then the lime for collecting information about the status of CE elements by agents of Areq block, interrogation by the Astate block and notifications of the Asm block will be the key for the resulting decision time.
Thus, the multiple of all messages of the MAMS node A: can be divided into a plurality ol external messages (the Areg agents receive from the CE element) and the internal messages (exchanged by the control blocks of the current MAMS node). The total decision time of the Asrrn block w ill consist of the time on receiving and forwarding internal and external messages. Denoting the multiple of all messages as C, the total message processing time by the Asm block will be determined by the expression (4), which will determine the value of the OF:
Cfe =T(c::;')+T(c;) (4)
All the multiple of external messages is determined by so many messages: messages of requests to the parameters of the CE element, messages of requests for the execution of failures commands (OAM) and the messages of notifications from the CE element. Then, the processing time of the external messages of the Aj MAMS node is determined by the expression (5).
TCC-T ) = T(C°'" ") + TIC°!" ") + TÍCT ua™)
(5)
Assume that some parameter j of ei element is requested with an average frequency vf and the time of data transfer from the
element to the managing agent of AKfA} block is T/' . l-'or
example, a CE network is defined, which consists of E elements and the management system consists of A agent nodes. Denote the number of diagnosed parameters inherent in the element et -PfeJ. The number of notifications from this element is N(et). The number of CAM control commands for this element is M(et). The number of elements entering the CE fragment under the management of the MAMS node Aj - E(AJ. Usually, OAM control commands are applied to the routes of the selected network fragment, which means that the number of such commands (requests for execution of commands) will be less than the number of elements of the CE fragment themselves. At the stage of planning the modeling of interaction processes, it is difficult to predict the ratio of the number of fragments, routes and elements of the network fragment in question. Therefore, in the calculations, OAM mechanisms are allowed for eaeh element, not the route. This will allow the calculations to get the upper limit of the OF for getting to an enlarged evaluation of the processing lime of messages in the interaction of MAMS and CE. Each MAMS node controls a CE fragment containing several network elements. The data transfer rate between the node A, and the e, element is determined to the minimum throughput value co( A., e,) = min e ) <&,- The frequency of requests node of agent At for the to the j-th parameter of the element c'/ - y. The volume of such a request
is vf*. The size of the response to this request is v*'*9 ■ The
predicate [BelfAj, pij] will say that the inversion of agent A, to the j-th parameter of the element et takes place. The predicate takes the value 1 in the case of truth, and 0 otherwise.
For all agents oiAngc block the total processing time for the parameters will be determined by the expression:
Ei J, I /=1
[_BelsoÂAn'\)\
T(C'') = T(C"-" ) + T(C'"-J ) + T(C"-s ) (7)
* A¡' v "staw "xttm Asnm'
In this paper, was selected a scheme for the internal interaction of control blocks and agents with the interrogation of the state and parameters of the element £4]. Such a scenario of interaction will increase the "independence" of both the agents themselves and the dynamism of the management system as a whole. Transformation expression (7) will be determined by the expression (8).
■» . A
EU)
T{C*)= X
"> '« III
* I
Ptr,}
l
/=1
•M
D(t,)(
+X
p^i-« j +
j
4* V<t \
{ y.pj
vf--v;
C0( ' ^STATEj )
\
lBelOUT(AREG¡,dlx)]+
TATE, » 4>ÏKr )
[BelabT(Asrm.,st)].
(8)
The total maximum time of the processing OF of all messages from agents and control blocks from the multiple A is determined by the expression (9).
^(C)=X(r_(C*)+T(G7)) (9)
i-i
To calculate the OF, the data presented in Table I were obtained:
Table 1
Initial data for calculation of OF
OAM,
(6)
The second term of expression (4) shows the time to process all internal messages of the MAMS node,The tasks of internal interaction of agents include the formation of requests for the necessary parameters for estimating the state (Astate ;) of each element and informing about the execution of control commands on the CE fragment. This is necessary to implement the process of controlling and managing failures on the network and further warning the Asm block about changes in the structure of the current CE fragment. Then the total time of interaction between blocks and agents inside the MAMS node will be determined by the expression (7),
Type of message Number Notes
Total number of requests to execute OAM commands 3 CCM, LBM, LTM
Interna! notifications 1 subject to damage to one route
External notifications 2 subject to damage to one route and / or fault indication
Duplicate messages from the agent of Ag^cj block 2 parameter data of the elements - under condition of serviceable routes and LTR -subject to damage to one route
The number of interrogation of the elements parameters 100 for the considered fragment of the network
Taking into account that all types of messages exist at the same time in the control circuit in question. Then the predicate values that enter the expression of the objective function will take the value of one. Thus, the estimation of the processing time of all messages transmitted between the CE fragment and the MAMS control blocks will be considered in the worst case (i.e., a preliminary aggregated estimate).
Simulation results
Based on the results of the calculations performed on this model, an initial estimate of the interaction of elements and agents.
The graph of the dependence in fig. 1 shows the influence of the frequency of requests for OAM commands on the control cycle time.
Fig. I. Time of implementation of the control cycle X(v) depending on the frequency v of sending messages (OAM commands): I, 2, 3 - I Tone, two and three faults in the route are detected, respectively; 4 - If there are six faults In the route
This allows us to draw the following conclusions: a high value of the control cycle time is traced at low frequencies of the CCM message sending (which indicates a large period between sending messages). At frequencies above 1-2 s_i, the control cycle time varies within a range of up to 200 s. It should be noted that this value is achieved with the number of detected route damage equal to several units (graphs 1, 2 and 3 in fig, 1). Graph 4 in fig. 1 illustrates the same result for the control cycle time, but if 6 or more faults are detected in the routes and at the sending frequencies of the CCM message from 6 s~l. Thus, the problem of the optimal choice of period for sending OAM command request is solved. This periodic can be determined with respect to mathematical expectations of failure rates.
The graph in fig. 2 shows the time dependence of the implementation of the control cycle on the number of nodes in the MAMS.
s X(v)
Fig. 2. The implementation time of the control cycle X(d), depending on the number of nodes ¿2 in the MSSU: 1, 2, 3- If one, two or three faults in the route are detected, respectively
It can be seen from the graphs that as the number of agent nodes in the MS increases, the total time increases sharply from a few minutes to several hours. This is due to the increase in the amount of transmitted information per one node of the control system. The obtained dependence allows estimating the necessary time resources for managing a network of a given scale, and also allows to correct the control cycle time by a reasonable choice of its computing resources.
Dependence of the implementation time of the control cycle on the capacity of message transmission channels is shown in lig. 3.
s X(v)
t \ ^ .__ 3
2 ...jC...
1
0 >10* JxIQ* 6x10* 8*1 o' I «10* bit's
Fig. 3. The implementation time of the control cycle X(Q), depending oil the capacity of the channels 0: 1, 2, 3 ■ if one, two and three faults in the route are detected, respectively; 4 - If there are six faults in the route
With a larger channel capacity, the total control cycle time is reduced, but to a marginal extent with respect to the entire network. The most optimal capacity is between 10 and 20 kbit/s, which corresponds to a value within 7 seconds of the control cycle time. The result obtained allows us to reasonably choose a range of capacity values for the transmission of control and managing information in accordance with the necessary time resource over the control cycle time.
sX(v)
Fig. 4. Time of implementation of the control cyc\*iX(V) depending on
the volume Vof the transmitted data; 1, 2, 3 - If one, two and three faults in the route are detected, respectively; 4 - [f there are six faults in
the route
S X(v)
0 20 40 SO SO 100
The dependence shown in fig. 4 shows the increase of the control cycle time with increasing the amount of data transferred. Such an increase is possible with increased requirements for accuracy, and a change in cycle time is possible with an increase in the amount of information measured. This in turn characterizes the delay of the packets in the diagnostic data transmission channels and demonstrates a slight increase in the total time required to process the transmitted messages (by 0,2 s) with the increase of one packet of information up to 10 bytes.
Conclusion
The above model for estimating the interaction between the MAC MS node and the CE network fragment gives an estimate of the value of the total message processing time, but does not allow for the probabilistic nature of the occurrence of faults.
The model parameters, the values of which were taken into account in the estimation of the objective function, significantly affect the cycle time of the process of managing and controlling the states of CE network elements. The presented multiple of parameters can be expanded to perform a deeper estimation, thereby demonstrating its flexibility and effectiveness in terms of initial data. Modern monitoring and diagnostic systems follow the path of continuous growth in the accuracy, reliability and completeness of the reflection of the state of the elements, which leads to a sharp increase in the volume of transmitted diagnostic data and an increase in transmission delays. So it leads to the entire control cycle time.
The model makes it possible to estimate the time process of control and management of the CE network elements on the basis of OAM mechanisms. The result of the estimate allows us to formulate a number of requirements for individual processes and the entire MS as a whole. In addition, the model makes it possible to solve the problem of a reasonable choice of the required values of the considered parameters for the transmission of diagnostic messages.
The developed and described model of the process of interaction between the MAMS node and the CE fragment can be the basis for forming the methodology for building a multi-agent
network management system for a new generation of CE. At the heart of the presented model lie the key algorithms (processes) for monitoring and managing the states of network elements.
The received results of modeling allow estimating the time of processing messages in MAMS depending on various characteristics of a technical subsystem. The result obtained in the course of this simulation can be the starting point for further simulation of the processes of periodic managing and controlling of the states of network elements in order to estimate the control cycle time and the correspondence of these indicators to the required values.
Refe lenses
1. Login, E.V., Anufrenko, A.V., Kanaev, A.K. (2017). Multi-agent approach to structure formation of management of transport networks based on Carrier Ethernet technology. Collection of scientific papers by 6th international conference on advanced infotelecommunication, vol. 2, pp. 57-59.
2. Beneta, E.V., Kanaev, A.K. (2016). Development of control algorithm by failures in the telecommunication network based on carrier ethernet technology. Proc, Section "Information technologies at transport" of Anniversary XV St. Petersburg int. Conf. ''Regional informât-ics-2016", pp. 95-100.
3. Login, E.V., Kanaev, A.K., Muravtsov, A.A (2017), "Carrier Ethernet network operation scenario and the assessment of control cycle duration". Bulletin of scientific research results [Electronic], no. 3, pp. 159-170, available at: http://bmi.info/yiew/BLmycK-24.html#/l58 (accessed 1 Jan 2018).
4. Vasilev, N.V. (2009). Models and method for constructing multiagent decision support systems for managing distributed objects. Abstract of Ph.D. Dissertation, System analysis, management and information processing (by industry). Saint Petersburg Electrotechnical University "LET!", Saint-Petersburg, Russia.
5. Ivanov, A.J. (2008). Mobile distributed databases of automated information management systems EMERCOM of Russia. Abstract of Ph.D. Dissertation, Information systems and processes, Saint-Petersburg University of State Fire Service of Emercom of Russia, Saint-Petersburg, Russia.
T-Comm Vol.12. #1-2018
МОДЕЛЬ ПРОЦЕССА ВЗАИМОДЕЙСТВИЯ СИСТЕМЫ УПРАВЛЕНИЯ С ЭЛЕМЕНТАМИ СЕТИ CARRIER ETHERNET
Логин Элина Валерьевна,
Петербургский государственный университет путей сообщения Императора Александра I ФГБОУ ВО ПГУПС, Санкт-Петербург, elinabeneta@yandex.ru
Канаев Андрей Константинович,
Петербургский государственный университет путей сообщения Императора Александра I ФГБОУ ВО ПГУПС, Санкт-Петербург, kanaevak@mail.ru
Дннотация
Carrier Ethernet (СЕ) является технологией построения сети нового поколения. Эта технология отличается высокой масштабируемостью и сложностью выполняемых процессов. К таким процессам относится комплекс реализуемых в рамках технологии СЕ механизмов по контролю и управлению как состоянием элементов сети, так и ее конфигурацией. Механизмы управления и контроля состоянием элементов сети СЕ (механизмы ОАМ) на сегодняшний день не имеют формализованного описания применительно к системе управления (СУ). Для решения этой задачи необходима разработка модели мультиагентной СУ сетью Carrier Ethernet. Что в свою очередь невозможно без наличия комплекса моделей, демонстрирующих ряд процессов при взаимодействии элементов сети и компонентов мультиагентной СУ. Разработанная модель процесса взаимодействия системы управления с элементами транспортной сети СЕ получена с использованием алгоритма последовательного оценивания, а результат моделирования продемонстрирован с помощью средств вычислительной среды MathCAD. В основе ключевого этапа моделирования, а именно, формирование множества типов сообщений для передачи данных управления, лежит сформированная ранее функциональная модель мультиагентной СУ транспортной сетью СЕ. Исходные данные для расчета целевой функции модели содержит набор подпроцессов, входящих в ранее разработанные алгоритмы контроля и управления состоянием элементов транспортной сети СЕ. Результаты исследования представлены в виде зависимостей длительности цикла управления в рамках процесса контроля и управления состоянием сетевых элементов от различных параметров передачи управляющей информации и структурных параметров системы управления. Обоснованно выбранные значения рассмотренных параметров передачи диагностических сообщений позволяют сформировать требования к отдельным процессам и всей СУ в целом. Результаты представленной в статье модели могут являться базовыми данными при имитационном моделировании отдельных процессов контроля и управления состоянием элементов транспортной сети СЕ, а также основой для формирования методики построения мультиагентной системы управления транспортной сетью СЕ.
Ключевые слова: сотвг Ethernet, агенты, многоагентные системы, мультиагентные системы, алгоритм последовательной оценки, среда MathCAD.
Литература
1. Логин Э.В., Ануфренко А.В., Канаев А.К. Мультиагентный подход к формированию структуры системы управления транспортной сетью связи на основе технологии Carrier Ethernet // Актуальные проблемы инфотелекоммуникаций а науке и образовании: сб. научн. ст. СПб.: СПбГУТ, 2017. Т. 2. С. 57-59.
2. Бенета Э.В., Канаев А.К. Формирование алгоритма управления отказами в телекоммуникационной сети связи, построенной по технологии Carrier Ethernet // Информационные технологии на транспорте: сб. материалов секции "Информационные технологии на транспорте" юбилейной XV Санкт-Петербург. Междунар. конф. "Региональная информатика - 2016", Санкт-Петербург, 26-28 окт. 2016 г. СПб.: ФГБОУ ВО ПГУПС, 2016. С. 95-100
3. Логин Э.В., Канаев, А.К. Сахарова М.А., Муравцов А.А. Сценарий управления сетью операторского класса Carrier Ethernet и оценка длительности цикла управления // Бюллетень результатов научных исследований. 2017. Вып. 3. С. 159-170 Электронный ресурс URL: http://brni.info/view/ выпуск-24.1ит!#/!58 (дата обращения 18.01.2018).
4. Васильев Н.В. Модели и метод построения мультиагентных систем поддержки принятия решений для управления распределенными объектами: дисс. канд. техн. наук. СПб.: СПб ГЭТУ "ЛЭТИ", 2009. 179 с.
5. Иванов А.Ю. Мобильные распределенные базы данных автоматизированных информационно-управляющих систем МЧС России. СПб.: СПбУГПС МЧС МЧС России, 2008. 152 с.
Информация об авторах:
Логин Элина Валерьевна, Петербургский государственный университет путей сообщения Императора Александра I ФГБОУ ВО ПГУПС ассистент кафедры "Электрическая связь", Санкт-Петербург, Россия
Канаев Андрей Константинович, Профессор, д.т.н., Петербургский государственный университет путей сообщения Императора Александра I ФГБОУ ВО ПГУПС заведующий кафедрой "Электрическая связь", Санкт-Петербург, Россия
