=Paper=
{{Paper
|id=Vol-2623/paper8
|storemode=property
|title=Intelligent Modeling of Unified Communications Systems Using Artificial Neural Networks (short paper)
|pdfUrl=https://ceur-ws.org/Vol-2623/short8.pdf
|volume=Vol-2623
|authors=Oleksii Ivanov,Liudmyla Koretska,Volodymyr Lytvynenko
|dblpUrl=https://dblp.org/rec/conf/intelitsis/IvanovKL20
}}
==Intelligent Modeling of Unified Communications Systems Using Artificial Neural Networks (short paper)==
https://ceur-ws.org/Vol-2623/short8.pdf
Intelligent Modeling of Unified Communications Systems
Using Artificial Neural Networks
Oleksii Ivanov 1[0000-0001-6119-4134], Liudmyla Koretska 2[0000-0002-4284-4936]
and Volodymyr Lytvynenko 3[0000-0002-1536-5542]
1, 2 Khmelnytskyi National University, Khmelnytskyi, Ukraine
3 Kherson National Technical University, Kherson, Ukraine
1ivanovov@ukrtelecom.ua
2 sorokinaluda2908@gmail.com
3 immun56@gmail.com
Abstract. A Unified Communications System (UCS) is defined by a process
that combines all means of communication into a universal communication sys-
tem that enables reliable connection of the system’s users at any time and place
to exchange information. The purpose of the UCS is to simplify business pro-
cesses by facilitating communication between people. A characteristic feature
of the UCS is its ability to allow two or more users to use several ways to
communicate and transmit information. Unified service enables every employee
to perform their duties better and faster, and companies - to change their busi-
ness processes, making them more efficient and optimized to meet customer
expectations and market demands. This increases the efficiency of the business
processes of a company of any size and at the same time reduces the total cost
of owning a communication infrastructure, which gives the enterprise an addi-
tional advantage over its competitors. The proposed model of the mechanism of
optimization of the information exchange methods for UCS provides the im-
plementation of the selection mechanism of the optimal IEM for specific sub-
scribers in a certain time period given the real distribution of methods of infor-
mation exchange between specific subscribers. This will minimize unproductive
waste of working time and significantly increase the ergonomics of the techno-
logical process. The model and algorithm for implementing the above mecha-
nism form the basis of developing an intelligent decision support system for de-
fining telecommunication service delivery strategies.
Keywords: Unified Communications System (UCS), Unified Service, Mecha-
nism of Optimization of the Information Exchange Methods., Artificial Neural
Network, Decision Support System (DSS), Information Exchange Methods
(IEM).
1 Introduction
To date, unified communications are represented by hardware and software products
based on the following major components [1-4]:
Copyright © 2020 for this paper by its authors. Use permitted under Creative Commons
License Attribution 4.0 International (CC BY 4.0). IntelITSIS-2020
• IP-telephony - integrated solutions allow to set up a modern telephone network,
spanning from multiple subscribers within a small enterprise (SOHO) or remote of-
fice to up to several hundred thousand subscribers in a distributed network of a
large corporation.
• Call Center - a hardware and software complex designed to automate and improve
the efficiency of processing a large number of calls coming from customers, part-
ners, and others, whether by phone, e-mail or company website.
• Video conferencing - a way of interaction that enables effective communication
between the central office and regional units, in the media, for setting up distance
learning, etc.
• Teamwork tools - multimedia conferences, conference calls, situation centers, etc.
They constitute a single system of setting up and running audio, video and web
conferences that use a web browser to collaborate on documents (in particular, in
the case of distance learning, slideshows, presentations and sharing other educa-
tional materials), editing files, text chat, whiteboarding, etc.
• Unified messaging system - includes the function of voice messaging (voicemail),
listening to your e-mail on your phone, ability to check voicemail via the Internet,
send, receive and forward fax documents.
• Mobility - allows you to use a single phone number, providing call routing to the
device, which makes it most convenient for the user to talk at any time - using an
office, home or mobile phone. Combined GSM / Wi-Fi phone devices allow em-
ployees to stay connected both in the office premises and wherever mobile cover-
age is available, providing switching between networks.
• Presence control - a service that uses dynamic location information and allows
users to check their colleagues' availability in real time and connect with them
quickly with the most convenient means of communication at the moment.
The evolution of unified communications is represented on Fig. 1 [3].
Fig. 1. Evolution of unified communications [3]
To date, the design and realization of the intelligent decision support system for de-
fining telecommunication service delivery strategies is the actual task for Ukraine.
2 Literature Review
Let's analyze the literature in search of known models, methods, tools for defining
telecommunication service delivery strategies.
In telecommunications, complex service systems rapidly develop. Their aim is
producing ICT services. Increased complexity is based on the convergence of industry
information technologies, telecommunications and media. Operators of the telecom-
munication network can modify their business strategies: they can no longer produce
ICT services in a vertically integrated way but must sell previous services as provid-
ers to other ICT service providers. Modular service concepts, which are known from
Service Science and IP research, can be used for this task. ICT service modules,
called Enabling Services, are provided on Service Delivery Platforms to support ser-
vice development [5].
Businesses are increasingly turning to operators as a single mechanism for net-
working solutions that go beyond simple communication from basic Internet and tele-
phone lines to cloud computing and more sophisticated networking features that may
be traditionally provided by IT professionals or system integration providers. With the
growing emphasis on new trends that consume data in the enterprise space, such as
M2M, IoT and cloud services, face the challenge of not only managing the delivery of
a new era of services but also creating new and innovative solutions capable of long-
term connectivity and 100% customer reliability [6].
Five strategies to improve customer experience in telecoms: employing omnichan-
nel support, implementing a customer-centric culture, deploying AI-based digital
tools, investing in visual engagement, going back to basics with the human touch.
IDC estimates that 75% of enterprise applications will use AI services by 2021 [7].
In [8] there is a strategy implemented by implementing actions aimed at maintain-
ing product quality to satisfy consumers, by developing innovative and diversified
products, as well as package offers, simplified tariffs and transparency bills, imple-
menting international relations, mainly in Latin America (Brazil and Argentina).
Adopting such a strategy required the operator to redesign the corporate organization,
making it more subtle and flexible, horizontally developing teamwork, dialogue and
successful interaction between management and operational staff to speed up the de-
cision-making process.
The traditional organizational structure of telecommunications companies is often
ill-equipped to overcome the significant challenges – such as the complexity of the
systems concerned and the associated cost pressures – posed by this changing envi-
ronment. Telecommunications companies can increase the efficiency of their opera-
tions and better synchronize the elements of the technological system with the help of
artificial intelligence components. The technological and organizational capabilities
of AI components allow telecommunications companies to build a strong foundation,
which allows them to offer enhanced innovation and enhanced customer experience,
which is now required by a new breed of consumers [9].
So, to date using of AI components for the UCS is an actual task. Let’s analyze the
literature in search of using of AI components for communications systems.
The paper [10] is a study of the application of Artificial Neural Networks for prob-
lem-solving and optimizing the performance in different communication systems.
Artificial neural networks are being used for spectrum sensing, pattern recognition
and many applications in wireless communication.
A number of current applications of neural networks to telecommunications are
summarized in [11] and some relevant topics for future research in this field provide.
In [12] a brief overview of signal recognition approaches is presented - classical
methods, new machine learning and deep learning schemes expand from modulation
recognition to wireless technology recognition with the continuous development of
wireless communication system.
In [13], for the first time in the field of optical camera communication systems, an
equalizer based on an artificial neural network (ANN) is used using an adaptive algo-
rithm to increase the data rate.
In the paper [14] the review of the use of machine learning in various spheres of
communications is given and two examples of application in wireless networks are
considered.
In [15] there is an approach of using ANN for predicting the quality of software
system for accounting and billing of the provision of services for access to the Inter-
net. In [16, 17] there is an approach of using ontologies for evaluating the sufficiency
of information in specifications of requirements of software system for accounting
and billing of the provision of services for access to the Internet.
So, the known models, methods, tools for defining telecommunication service de-
livery strategies don’t solve the task of developing an intelligent decision support
system for defining telecommunication service delivery strategies. The aim of this
research is of developing an intelligent decision support system for defining tele-
communication service delivery strategies.
3 A Decision Support System Based on Using the Most
Convenient IEM in a Certain Amount of Time
A characteristic feature of a modern UCS is the ability to use several methods of es-
tablishing communication and transmitting information between clients, regardless of
their location and the type of equipment used. This can be a combination of a landline
or cellphone, software on a smartphone or desktop computer, voice and email, fax and
instant messaging, multimedia conferencing, and more.
One way of utilising this capability of UCS is to use a system that would, at some
point in time, determine the optimal way for information exchange for a pair (or a
group) of subscribers. For example, when users are participating in a video or audio
conference, incoming phone calls are not desirable, and in some situations are not
allowed. The same applies to a lunch break when an employee is not sufficiently fo-
cused on the production process. Such a system should include an individual work
schedule of each user, based on which the selection of the optimal type of communi-
cation for the information exchange between certain users at a certain time would take
place.
Fig. 2 presents a schematic model that includes a mechanism to optimize the use
of information exchange methods (IEM) as the main component of this structure.
Although this model is assumed to be scalable, certain limitation is required for the
operation of the optimization mechanism. This limitation can be considered as a con-
stant value, for example, during the learning of the neural network (Tl).
Fig. 2. Schematic model of the mechanism of optimization of the information exchange methods
The model also defines the concept of "special users". These may be subscribers
with a distinct work schedule (general manager, internal security service, etc.). The
"XnXm" block determines the input vector for the optimization unit to function and,
together with the values of Y, generates a set of combinations of specific users and
corresponding AEs. The product of Xn Xm Yk is the output value and characterizes a
specific SIO for a given pair of subscribers at a particular moment in time.
4 Artificial Neural Network as "robot-assistant"
The task of finding the optimum method of information exchange is complex and
difficult to formalize, since the configuration of information exchange is usually hard
to predict and describe by the known mathematical methods used in telecommunica-
tions (Teletraffic Theory, Mass Service Systems (MSS), Markov Models, etc.).
The author proposes to solve this problem with artificial neural networks (ANN).
Fig. 3 presents a schematic simplified structure of a neural network
Fig. 3. ANN Structure
Fig. 3 shows the structure of a neural network for three subscribers
Х = { Х 1 , Х 2 , Х 3 } and three time intervals Т = { ∆t1 , ∆t 2 , ∆t 3 } , where: ∆t1 – 9.00-
11.00; ∆t 2 – 11.00-15.00; ∆t 3 – 15.00-18.00.
The architecture of the developed ANN corresponds to a rectilinear multilayer
perceptron. The input layer of the ANN reflects the set IEM noted above when con-
sidering the structural scheme for optimizing the provision of telecommunications
services. There are two hidden layers of neurons in ANN: Т = { ∆t1 , ∆t 2 , ∆t 3 } and
A = { A1 , A2 , A3 } .
The first one shows time intervals of a certain length and the second one is ap-
proximate and is intended to approximate a function. In general, both hidden layers
can be considered as one approximation. In this case, the accuracy of the approxima-
tion increases due to the increase in the number of neurons in the hidden layer.
Functionality of the output layer of ANN corresponds to the choice of the optimal
way of information exchange at a certain period of working time.
For the activation of neurons in the multilayer perceptron, we adopt sigmoid and
jump functions, which are most suitable for the ANN of this type.
ANN is trained via the widely described in the literature algorithms of perceptron
training using the error correction learning rule. Algorithm learning and functioning
of the network constitutes in the following:
─ A1.1. Assign values to weight coefficients W randomly or following the inputs
from the experts.
─ A1.2. Enter the input vector corresponding to the sets Х Х ,Y .
─ A1.3. Enter the values of the X ,X ,Y outputs of the training sample of the un-
trained network.
─ A1.4. Enter the number of iterations for the ANN training.
─ A1.5. Set the learning speed .
─ A1.6. Obtain weight coefficients .
─ A1.7. Introduce an input vector for the operation of the ANN and obtaining the
functionals.
─ A1.8. End of algorithm.
5 Conclusions
The proposed model of the mechanism of optimization of the information exchange
methods for UCS provides the implementation of the selection mechanism of the
optimal IEM for specific subscribers in a certain time period given the real distribu-
tion of methods of information exchange between specific subscribers. This will min-
imize unproductive waste of working time and significantly increase the ergonomics
of the technological process. The model and algorithm for implementing the above
mechanism form the basis of developing an intelligent decision support system for
defining telecommunication service delivery strategies.
References
1. Unified Communications System, 2012,
https://www.cisco.com/c/en/us/td/docs/solutions/Enterprise/Data_Center/VMDC/3-
0/UC/UCS.html, last accessed 2020/04/27.
2. Oracle® Communications UnifiedCommunications Suite, 2015,
https://docs.oracle.com/cd/E65115_01/doc/ucs_70_deploy_guide.pdf, last accessed
2020/04/27.
3. Unified communications (UC), 2020,
https://searchunifiedcommunications.techtarget.com/definition/unified-communications,
last accessed 2020/04/27.
4. Scott, R.: What is Unified Communications?, 2020, https://www.uctoday.com/unified-
communications/what-is-unified-communications/, last accessed 2020/04/27.
5. Wulf, J.: Analysis and design of value production strategies and business models in the
telecommunications industry. Berlin, Schriftenreihe Informations- und Kommu-
nikationsmanagement der Technischen Universitat Berlin (2012).
6. Next-generation enterprise service delivery – what do operators need?, 2015,
https://telecoms.com/wp-content/blogs.dir/1/files/2016/03/Amdocs_Whitepaper_Hi_res-
v31.pdf, last accessed 2020/04/27.
7. Mort, A.: Five strategies to improve customer experience in telecoms, 2019,
https://customerthink.com/five-strategies-to-improve-customer-experience-in-telecoms/,
last accessed 2020/04/27.
8. Scalera, F.: Development Strategies of Telecommunication Companies after the Markets’
Liberalisation. A Case Study: Telecom Italia. Procedia - Social and Behavioral Sciences.
58, 149 – 158 (2012).
9. Halaoui, H., Sammour, D., Zein, H.: IT and network integration in telecom companies.
Creating efficiency and customer satisfaction, 2013,
https://www.strategyand.pwc.com/m1/en/reports/it-and-network-integration-in-telecom-
companies.pdf, last accessed 2020/04/27.
10. Pattanayak, S., Das, P.: Artificial Neural Networks for Intelligent Communication Sys-
tems: A Study. International Journal of Engineering Research in Computer Science and
Engineering. 5 (2), 186-189 (2018).
11. Clarkson, T.: Applications of Neural Networks in Telecommunications. Neurocomputing.
10 (4), 415-417 (2011).
12. Li, X., Dong, F., Zhang, S., Guo, W.: A Survey on Deep Learning Techniques in Wireless
Signal Recognition, 2019, https://www.hindawi.com/journals/wcmc/2019/5629572/, last
accessed 2020/04/27.
13. Younus, O., Hassan, N., Ghassemlooy, Z., Haigh, P., Zvanovec, S., Alves, L., Minh, H.:
Data Rate Enhancement in Optical CameraCommunications Using an ArtificialNeural
Network Equaliser, 2020,
https://strathprints.strath.ac.uk/71997/1/Younus_etal_IEEEA_2020_Data_rate_enhanceme
nt_in_optical_camera_communications.pdf, last accessed 2020/04/27.
14. Samek, W., Stanczak, S., Wiegand, T. The convergence of machine learning and commu-
nications. ITU Journal: ICT Discoveries. 1 (1), 49-58 (2018).
15. Pomorova, O., Hovorushchenko, T.: Research of Artificial Neural Network's Component
of Software Quality Evaluation and Prediction Method. In: The 2011 IEEE 6-th Interna-
tional Conference on Intelligent Data Acquisition and Advanced Computing Systems:
Technology and Applications Proceedings. Prague (2011).
16. Pomorova, O., Hovorushchenko, T.: The Way to Detection of Software Emergent Proper-
ties. In: The 2015 IEEE 8-th International Conference on Intelligent Data Acquisition and
Advanced Computing Systems: Technology and Applications Proceedings. Warsaw
(2015).
17. Hovorushchenko, T., Pomorova, O.: Information technology of evaluating the sufficiency
of information on quality in the software requirements specifications. CEUR-WS. 2104,
555-570 (2018).