TWO APPROACHES FOR TELECOMMUNICATION NETWORKS MANAGEMENT
DOI 10.24411/2072-8735-2018-10053
Alexander B. Goldstein,
Bonch-Bruevich Federal Telecommunications University, St.Petersburg, Russia, [email protected]
Keywords: heterogeneous networks, business process management, self-organizing management, resource management, policy-based management.
Modern NGN/IMS and post-NGN networks are extremely complex and very hard to manage. It is both difficult to manage the networks according to new requirements, and to do it on control and data planes in real time. We present in this paper a new complete approach for these networks management models and methods. Traditional telecommunications networks management methods are oriented on fixed network architecture, stable traffic structure, bounded list of telecommunication services. New NGN/IMS and post-NGN networks requirements need to extend that models and methods with ideas and results from cognitive multi-agent models.
Networks are used in greater degree of provision of mobile subscribers, communication in social networks, the exchange of text messages, positioning, listening to music, viewing video clips, network games and many other entertainment applications in real time, multimedia conference communication, transfer of banking information and telemetry, Internet, the interaction of computers M2M, management in on-line mode, i.e. than for traditional telecommunications services. New services require fundamentally new and rapidly changing approaches to decision-making in the management of telecommunications Operator, a regular re-evaluation provided for the adoption of information from the OSS/BSS (Operation/Business Support systems- support systems of operations/business).
Для цитирования:
Гольдштейн А.Б. Подходы к управлению современными сетями // T-Comm: Телекоммуникации и транспорт. 2018. Том 12. №3. С. 57-63.
For citation:
Goldstein A.B. (2018). Two approaches for telecommunication networks management. T-Comm, vol. 12, no.3, pр. 57-63.
-e-
1. Introduction
Modern NGN/IMS networks are used in greater degree of provision of mobile subscribers, communication in social networks, the cxehangc of text messages, positioning, listening to music, viewing video clips, network games and many other entertainment applications in real time, multimedia conference communication, transfer of banking information and telemetry, Internet, the interaction of computers M2M, management in online mode, i.e. than for traditional telecommunications services [ I j. New sen.1 ices require fundamentally new and rapidly changing approaches to decision-making in the management of telecom 111 un i cat ions Operator, a regular re-evaluation provided for the adoption of information from the OSS/BSS (Operation/ Business Support systems-support systems of operations/ business) [2].
2. Evolution of Hie telecommunications
networks management
Dynamics of the change in telecommunications networks management presented in the fig. 1. Here we speak about two methods to control info communications. We can definite them like centralized fixed designing, and, otherwise, decentralized dynamic self-organiza! i on.
Both of these approaches were always presented in commu-nicational and computer networks managing in some proportion. Correlation between these methods is achieved a new way every time at next stage of info communication development. Modem stage of synergy of these both approaches to info communication management is represented in 3GPP documents, devoted to SON {Self-Organization network), and in TM Forum documents
I IS
■ Top-down Managorrofil Chain
The Mhtïie.irr or ihe
Fig. 1. Dynamics ofapproach changing to info communication managing
Development of Frameworx architecture includes the description of new requirements to managing and operational support due to the shift of info eommunieational operators business paradigm [2]. Exponential rising of number of info communication technologies using in modem telecommunication networks, and complication of logical interfaces between NCOSS and functional subsystems of business support BSS, subsystems of services and resources life cycle control system, subsystems of network services production, determination of basic business substances in info model NGOSS according to SID (Shared Information/Data) and dynamics of changing, complicated correla-
tion between business substances and functional areas of Frameworx - all these require to attract models and methods of self-organization and multi-agent systems of cognitive info communication managing.
Situation further develop trend is imaged in fig. I, according to authors opinion. Concerning the first traditional approach (fig.l), we note that it is still installed in model OS1, where the controlled objects are Identified by abstract expressions like MO (Managing Objccts) .Recommendations for determination of controlled objects GDMO (Guide lines for Definition of Managed Objects) expand the language of abstract notifications from recommendation ITU-T Z.105 specially to provide a method for controlled objccts definition. We emphasize that control and controlled objects, called by mangers and agents, respectively, are considered as equal code. Further, during the working-out the Internet Network Management Standards, the Working Group IETF received the development of the prototype of Count Simple Network Management Protocol (SNMP), which through the Intermediate versions has developed into version of 3 and defined the architecture for objects of the manager and the agent, the communication protocol, and also the well done Management Information Base (MIB) with definitions for most of the items in the IP network. The same time when the ITF were elaborating the simplified technology of' IP networks management, telecommunication companies were to control large, complex networks with special requirements of alacrity with five nines and ability to unite operators for providing new Info communication services.
Therefore, ITU-T developed a set of standards Telecommunication Management Network that suit the specific requirements for networks of communication and formed the base for further development of the TM Forum [2].
3, Decision support multi-agent model
for telecommunications network management
In order to support decision making in telecommunication network management and, in particular, in order to produce search and processing B1 data in shared IT-landscape of Telecommunication Operator, it is most effective and expedient to use the multi-agent approach [3, 4], in frame of which the system is constructed as a set of agents (Agents BJ, Agents of Manager, Agents of OSS/BSS-subsystcms). The choice of multi-agent technology makes it easy to combine in united system general purpose protocol of codes and each of the other tools for the work with specific data base types of managing subsystems and info communications. But before discussing the formalized mathematical model and the research itself, we will consider the physical model of the telecommunications network management, which is presented in fig.2.
As it is shown in fig. 2, the basic Bl operations (Business Intelligence) include the sending requests to various data of the technical accounting systems of NRl (Network Resources Inventory), billing, technical maintenance (Fulfillment), Performance Analysis Security, FM (Fraud Management), control quality of SQM (Service Quality Management), etc., as well as collection and Integral processing of the results of these interviews of sources, presented in fig. 2. Wherein the results obtained to the same request, but from other systems, as a rule, should supplement each other. It is also possible that obtained from different systems information will be duplicated or differentiated by the degree.
58
-e-
ir
Fig, 2. Physical model
The further evolution of the development of NGN/IMS networks, the creation of new services, the shift in the traffic to the side of [oT, development services related to the positioning of subscribers, diversity network elements and the introduction of the principles of self-organization in building communication networks evolve - these problems of BI information processing when managing super-complex post-NGN networks, become more significant. Therefore, multi-agent systems (MAS) are a suitable basis for resolving the problems with the construction of this formalized model, which enable us to evaluate the reliability, performance and efficiency of the control system. [5] The article also proposes a new multi-agent concession method for processing information in the Bi. presented below. Suppose that the information for decision-making by 1) telecommunications management system consists of N portions
lï= {D„D,.....Du S,
(1)
which are located in N systems comprised lo network manage complex (inventory, billing, fault management, quality control, performance analysis, fraud-management, etc.)
ç= {S:, S3.....SN}
(2).
At the same time it has a place and is allowed, as it has been marked above, some overlapping of data.
Di n Dj * 0
(3)
Fig. 3 shows the multi-agent system from N + 1 agents
M = {Ml, M2.....MN+1}, corresponding to the physical
model on iig.2. Here, each agent implements his own strategy ii.Agents have access to various control systems and return information on the topic of the request from BI. For (he request with index J the agent Mi sends data Dj" from the collection of data Dj. Further, the data sent by the agents is combined (Integrate) in two ways: data are combined obtained from one system i, and then the data from all systems i=l,2...
**(n,, % («,, (at. Dp) -m l4)
+
To assess the quality of the information D collected by multi-agent system in lig.3 and thus combined we use the criteria known from the theory of searching. Actually for the aims of this study only two parameters are important: the fullness (nothing Is forgotten needing for Rl) and the accuracy (nothing unnecessary to be send). These parameters often refer lo a one-word relevance (that means suitable, relating to matter).
Here Relevance is the accordance between the answer and the request, accounting such concepts as fullness and accuracy. Fullness is the ratio of the number of received results to the total number of data existing in the operate info communication system relevant to this request (^.-queries for the intensity ).).
p =
D~
(5)
Accuracy is the ratio of the number of relevant results to the total number of results in response lo this request.
R =
(6)
Fig. 3, Multi-agent model
For estimating the effectiveness of the information searching other criteria is often used: the coefficient of loss information and the coefficient of searching noise. In our study, for obvious reasons, it is assumed that the coefficient of loss of information = 0, the coefficient search noise = 1. The query/response loss rate between the OSS subsystems is 0, which is quite natural for any industrial OSS/BSS system. The accuracy can fluctuate in the range 0.S-1.0, the fullness is usually close to the values of 0.80.9. It is much more important that for the tasks of this study these criteria of completeness and accuracy are necessary apply a slightly complicated version.
The fact is that for multi-agent system cffcctivc decision making to operate info communicate network, as noted above, we need integrated processing of data requested from vary systems OSS/BSS. At that to ensure the effectiveness of these operate decisions, there arc required not only accuracy and fullness of data response from vary systems OSS / BSS, but also, first of all, the accuracy and completeness of the corresponding values of the integral function FDi. To achieve that we need a simultaneous, cooperative processing of data produced by agents in different systems, with appropriate "equalization" of time delays, for it is proposed to fill out the source-agents representing the corresponding from operational management subsystems, periodically
-е-
Step 5. Assign NB = NB/2; N = NA + NB. Calculate T <lc> by formula (4). Check the fulfillment of the inequality I - (t) < i,. If the inequality is satisfied, put NA = N, otherwise, N = NA.
Step 6, Cheek the condition NB = 1. If the condition is not fulfilled, go back to step 5. When the condition is met, read the current value is selected and complete the algorithm.
It should be noted that, despite the assumption of exponential distribution of gaps between points signai arrival, giving a somewhat pessimistic estimate admissible contribution of the Erlang distribution, adequacy of the proposed mode! is confirmed by a number of experimental dala. In the final section of the article, let us consider the development of formulated Lhe multi-agent approach to the investigation process management of communications on a real scale time. In other words, here the strategy of lhe reflective agent is mapping from lhe current network slates lo appropriate action because in many situations it is permissible consider thai the stale of the data at the time indicates a complete description history to the time t.
For such a managed object state, which prov ides all the important information about the past, they say that it is Markovian or possesses the property of Markov, From the above, we can draw a conclusion lhat in Markov world agent can safely use a strategy without memory for decision making instead of a theoretically optimal strategy, which may require a greater memory size-
So, we have considered how an agenl strategy can depend on lhe last event and individual characteristics of environment. I lowever, as we discussed at the beginning, the adoption of optimal solution can be deduced from lhe assessment of the future. To complete this section il is important to emphasize that proposed in this chapter mathematical methods of organization operational management of the Communications Operator facilitates the transition of a promising client-cenlric model (hereof in more details in Chapier 5), because besides all aboul ihese methods provides increase in demand for info communication services on lhe simple reason thai a clieni-centric model is a huge amount of information about each subscriber and devices connected to the Operators network, therewith information in realtime mode,
5. Management elements tor telecommunication operator
According the section 1, conception of the self-organizing networks (SON) was created by a non-profit foundation 3rd Generation Partnership Project (3GPP) with purpose to increase efficiency and flexibility of telecommunication networks management, reducing the cost and increasing profit of the Operator, and was originally oriented lo the application lhe mobile communication networks of generaiions 3G /4G/5G. There is defined lhe key position in recommendations 3GPP at conception SON: concerning its implementation, there is said that SON and associated with ils applying the benefits are extremely important and extremely demanded by the telecommunicate Operators.
Consequently, a modem and forward-looking network management system should fully support the functionality of the SON, as defined in 3GPP standards [8].
Simultaneously, the conception was developing and is developing in the working drafts of the TM Forum noil-commercial organization that is de facto leader in management and automation communication enterprises.
The ideas developed by TM Forum connect IT developers, vendors of telecommunication equipment, communication Operators and are fair in relation to management any mobile network, such as a fixed connection. These ideas have determined the modern approach to automation of network processes, assuming the presence of an IT landscape Operator for OSS-complex. hi department of the network operation the complex should include a system of technical accounting (Network Resource Inventory System (NRI), which is lhe foundation for work w ith physical and logical rcsourccs, and is a base for interacting equipment system, that is to manage network objects, including the possibility of automatic provision of data to lhe NR1 system, testing equipment, survey network, the activation of equipment (services). Such daia exchange is organized using iniemiediale software (PSP), or Middleware, by the way of applying standard interfaces for protocols, MTNM, SNMP, MTOS1 etc., and/or specially developed command codes.
According the management tasks in mobile networks, we mean, first of all, the network manage systems of NMS (Network Management System) type, i.e. program-hardware tools that help the IT specialist of the Operator to realize:
1) Network research (Network device discovery) with purpose of detection of new netw ork elements;
2) Monitoring network devices (Network device monitoring) on matter of devices good condition and compliance with the SLA;
3) Network performance analysis, monitoring key indicators such as uptime of equipment, delay and loss ofihe package;
4) Set up sman notifications-customizable alerts that will respond to specific network scenarios by an electronic mail or telephone.
Concerning the 4G/5G networks, they have in a view, by tradition, the mobile part of lhe network only, as it was while 2.5G/3G existed, when a fixed part of the network was living its own life and was operated apart. However in today's reality of lhe generation of 4G/5G, as point of view at services, and review of lhe operational control tasks, it is not the besl sirategy to separate lhe tasks of managing transport networks and equipment.
The main reasons for the introduction of SON architecture is the natural desire of the Operators to abridge (at least not increase) operating expenses by the reduction of the human intervention degree in the implementation planning, implementation and operation of their networks, reduce capital costs by optimize the use of their network resources, save and to increase profits, reducing the errors from the human factor. In the age of high-growth urban mobility broadband intemel access, have a place al regional Russian market in the present time is rapidly increasing demand for exchanging data service, provoking the development of technologies radio department of chain. The independent analysts point oui thai volumes of operational lasks, their quality and speed Implementation will be differeni in lhe near future from present. Consequently ii concerns planning and deployment processes of the operator's network while it is expanding, to ensure the required quality of service and uninterrupted work of the whole network infrastructure.
Operational management of OAM & P is play ing an important role in achieving these ihrce goals, uniting resource management (Operations), administration management (Administration), technical maintenance (Maintenance) and new resources input (Provisioning). !f these lasks are directly are controlled by
61
personnel, even with automation tools, slill this manual labor requires a lot of time, significant financial costs, highly qualified specialists, and the human errors are equally possible. Therefore, the principles embodied in the concept of self-organizing networks, begin to be reflected and applied in reality existing in operators networks.
New method and mathematical models presented above in fig, 3 and fig. 4 allow to assess in practical plan the opportunities and boundaries of SON functionality in terms of interaction with OSS/BSS systems and technologies in the structure of automation systems, and answer the question, how the described technology can approximate operators to cherished goals. To ensure end-to-end business processes of the operator need the interaction of the OSS/BSS systems with network elements and a network management system. MW systems are intermediate software between equipment (and their systems management) and high-level applications of class OSS/BSS. Consequently, middleware systems will be a level - Intermediary between SON functions, that is, the network level, and high-level BSS/OSS applications, such as Inventory, CRM, billing frame, as represented in this fig, 6,
data" are determined by one middleware function. The processes of the remaining four groups sufficiently tightly use the functions of middleware; especially it concerns the processes of the group "Provision of services resources." From position of inventory and technical accounting the operational processes implemented by mechanisms and algorithms of SON in the part of self-configuring and optimizing network elements infrastructure Operator are important for automation,
6. Conclusion
This article is a part of the research for perspective networks management modeling, formally specifying and developing post-NGN management methods. The model described above allows post-NGN operators to abstraeL physical network resources, to add in/replace them with software-defined and virtualized network functions, to orchestrate resource provisioning in an automated way with multiagent system on fig, 2 and 3.
References
Customer
-e-
\ CRM /
\ /
NR1 \ / Billing
\ /
S f^dle^ re
Fig. 6. Shock "passing" of the business process through the system operator (NE Network Llement, EM-Element Manager, NM-Network Manager)
According the map of eTOM and define the groups of processes, w hich can be attributed to the functions of middleware. The groups of processes "Provision and support of the resource level processes" and "Collecting and spreading the resource
1. Goldstein 13,, Koucheryavy A. (2013). Post-NGN era communication networks. BHV-Petersburg.
2. Samuylov K„Chukarin A„ Yarkina N. (2015). Business processes und information technologies in management of telecommunications companies, Alpina Publish.
3. Gorodetsky V.l., Skobelev P.O. (2017). Industrial Applications of Multi-agent Technology: Reality and Perspectives. SP!IRAS Proceedings.
4. Weiss Gerhard (2013). Multiagent Systems (Intelligent Robotics and Autonomous Agents series). The MIT Press; second edition.
5. Ong Jason, Abidi Syed (1999). Data Mining Using Self-Organising Ko honen maps. A Technique for Effective Data Clustering & Visualization International Conference on Artificial Intelligence (IC-A1'99), Las Vegas, June 28-July I 1999.
6. Gelenbe E., Mitrani 1. (2010). Analysis and Synthesis of Computer Systems. World Scientific, Imperial College Press, London and Singapore.
7. Jung M.M. (1991). Busy period distribution in an SPC processor having a cfock_pulse operated gale. Philips Telecommunications. Vol.9. No. 2.
8. GB92')D_ A |ip Ileal to ti_!;ramcwork_R 14.5.0, [Llectronic Resource]. Access Mode: www.tmforum.org.
9. Krief, F. (2006). Réseaux Autonomiques. 7ème Colloque Francophone of Gestion of Réseattx et of Services. Bordeaux, France. LaBRI, Université Bordeaux-!.
10. Riedl, A. (2002). Hybrid Genetic Algorithm goes Routing Optimization in IP Networks Utilizing Bandwidth and Delay Metrics. IEEE Network Magazine, September 2002.
-e-
ii
ПОДХОДЫ К УПРАВЛЕНИЮ СОВРЕМЕННЫМИ СЕТЯМИ
Гольдштейн Александр Борисович, к.т.н., доцент, Санкт-Петербургский государственный университет телекоммуникаций им. М.А. Бонч-Бруевича, Санкт-Петербург, Россия, [email protected]
Aннотация
Современные инфотелекоммуникационные сети NGN/IMS и post-NGN крайне сложны с точки зрения управления. Появляются новые требования, новые процессы управления в реальном времени. Целью данной работы является предложение новых моделей и методов для управления бизнес-процессами на современных сетях, с применением когнитивных многоагентных систем.
Ключевые слова: гетерогенные сети, управление бизнес-процессами, самоорганизующиеся сети, управление ресурсами, управление сетью. Литература
1. Гольдштейн Б.С., Кучерявый А.Е. Сети связи пост-NGN. СПб: БХВ-Петербург, 2013.
2. Самуйлов К.Е., Чукарин А.В., Яркина Н.В. Бизнес-процессы и информационные технологии в управлении современной инфокоммуникационной компанией. М.: Альпина Паблишерз, 2015.
3. Городецкий В.И., Скобелев П.О. Многоагентные технологии для индустриальных приложений: реальность и перспектива / Труды СПИИРАН. 2017.
4. Weiss Gerhard. Multiagent Systems (Intelligent Robotics and Autonomous Agents series). The MIT Press; second edition, 2013.
5. Ong Jason, Abidi Syed. Data Mining Using Self-Organizing Kohonen maps: A Technique for Effective Data Clustering & Visualization International Conference on Artificial Intelligence (IC-AI'99), Las Vegas, June 28-July 1, 1999.
6. Gelenbe E., Mitrani I. Analysis and Synthesis of Computer Systems. World Scientific, Imperial College Press, London and Singapore, 2010.
7. Jung M.M. Busy period distribution in an SPC processor having a clock_pulse operated gate. Philips Telecommunications, 1991. Vol.9. No.2.
8. GB929D_Application_Framework_RI4.5.0. [Electronic Resource]. Access Mode: www.tmforum.org.
9. Krief F. "Reseaux Autonomiques." In: 7eme Colloque Francophone of Gestion of Re?seaux et of Services, 2006, Bordeaux. France. LaBRI, Universite? Bordeaux-I, 2006.
10. Riedl A. Hybrid Genetic Algorithm goes Routing Optimization in IP Networks Utilizing Bandwidth and Delay Metrics. IEEE Network Magazine, September 2002.
Информация об авторе:
Гольдштейн Александр Борисович, к.т.н., доцент, Санкт-Петербургский государственный университет телекоммуникаций им. М.А. Бонч-Бруевича, Санкт-Петербург, Россия