Page 25 - Proceedings of the 2018 ITU Kaleidoscope
P. 25
A MACHINE LEARNING MANAGEMENT MODEL FOR QoE ENHANCEMENT IN
NEXT-GENERATION WIRELESS ECOSYSTEMS
3
1
2
Eva Ibarrola , Mark Davis , Camille Voisin , Ciara Close , Leire Cristobo 1
3
1 University of the Basque Country (UPV/EHU), Spain
2 Dublin Institute of Technology (DIT), Ireland
3 OptiWi-fi, Ireland
ABSTRACT access ensuring the coverage, mobility and accessibility
demanded by their users. Consequently, many scientific and
Next-generation wireless ecosystems are expected to industrial researchers [2] envisage 5G as having an agnostic
comprise heterogeneous technologies and diverse radio access network (RAN) comprising multiple wireless
deployment scenarios. Ensuring a good quality of service technologies.
(QoS) will be one of the major challenges of next-
generation wireless systems on account of a variety of Therefore, even though there is still no clear consensus
factors that are beyond the control of network and service about what the next-generation wireless (NGW) era will
providers. In this context, ITU-T is working on updating the embrace, there seems to be a general agreement that 5G
various Recommendations related to QoS and users' quality will comprise heterogeneous networks (HetNet)
of experience (QoE). Considering the ITU-T QoS cooperating to maintain a user's QoE. Furthermore, the
framework, we propose a methodology to develop a global adoption of new business models, quality of business
QoS management model for next-generation wireless (QoBiz), to integrate all the capabilities that
ecosystems taking advantage of big data and machine next-generation wireless systems may offer will be crucial.
learning. The results from a case study conducted to Nonetheless, ensuring the required quality of experience in
validate the model in real-world Wi-Fi deployment these complex scenarios will become a major issue.
scenarios are also presented.
The ITU’s standardization expert group for future networks
Keywords – Big data, machine learning, QoBiz, QoE, QoS, (SG-13) has been working towards the definition of 5G
Wi-Fi systems and the development of new Recommendations
related to QoS in the NGW environment [3, 4]. In addition,
1. INTRODUCTION being aware of the great challenge of managing
next-generation wireless networks, new groups, like the
The evolution of Internet users' behavior in recent years, “Focus Group on Machine Learning for Future Networks
along with the increasing variety of free applications and including 5G” [5], have been established. ITU-T SG-12 has
services, has led to Internet access becoming something also focused on updating and defining new
indispensable for our daily life. As a result, users are Recommendations related to QoS and QoE for adapting to
turning out to be more and more demanding in terms of the new NGW scenario [6, 7]. Nevertheless, there is still a
Internet coverage, accessibility and mobility, causing need for methodologies that will take advantage of new
Internet service providers (ISPs) to consider alternative techniques and mechanisms, such as machine learning (ML)
business models to fulfill these needs. For this reason, some algorithms, for the deployment of QoS management models
of the technologies that were originally considered for as defined in the standardized QoS frameworks.
providing local access to the Internet have now emerged as
ubiquitous access technologies, leading to complex In this paper, a methodology for the implementation of a
scenarios where fulfilling the required quality of service global QoS management model in NGW ecosystems is
(QoS) will become a real challenge. proposed. The model takes into account the ITU QoS
framework [8] and the methodology aims to include all the
The Wi-Fi technology, defined in the IEEE 802.11 standard, aspects that the 5G era will require. Enhancing the quality
is a good example of this. While originally designed to be a of experience and the satisfaction of the users through new
wireless local area network (WLAN) technology, today's business models to fulfill their requirements are the main
large deployment of Wi-Fi networks has encouraged ISPs target of the methodology. The identification of the optimal
to consider new business models turning this technology key performance indicators (KPIs) and key quality
into a “ubiquitous access technology for mobile users” [1]. indicators (KQIs) are essential to achieve this goal.
In this way, providers can offer complementary broadband
978-92-61-26921-0/CFP1868P-ART @ 2018 ITU – 9 – Kaleidoscope