The Role and Applications of Machine Learning in Future Self-Organizing Cellular Networks

The Role and Applications of Machine Learning in Future Self-Organizing Cellular Networks

Paulo Valente Klaine (University of Glasgow, UK), Oluwakayode Onireti (University of Glasgow, UK), Richard Demo Souza (Federal University of Santa Catarina (UFSC), Brazil) and Muhammad Ali Imran (University of Glasgow, UK)
DOI: 10.4018/978-1-5225-7458-3.ch001

Abstract

In this chapter, a brief overview of the role and applications of machine learning (ML) algorithms in future wireless cellular networks is presented, more specifically, in the context of self-organizing networks (SONs). SON is a promising and innovative concept, in which future networks are expected to analyze and use historical data in order to improve and adapt themselves to the network conditions. For this to be possible, however, algorithms that are capable of extracting patterns from data and learn from previous actions are necessary. This chapter highlights the utilization and possible applications of ML algorithms in future cellular networks. A brief introduction of ML and SON is presented, followed by an analysis of current state of the art solutions involving ML in SON. Lastly, guidelines on the utilization of intelligent algorithms in SON and future research trends in the area are highlighted and conclusions are drawn.
Chapter Preview
Top

Introduction

In the last couple of years, mobile wireless networks have become an essential part of our lives, due to a broad range of applications and services that are available. Nowadays, people are able to do business on the go, by performing teleconferences whenever and wherever needed, watch or listen to their favorite videos and music on the fly, talk to distant relatives, stream audio/video whenever a special event happens, instantly upload photos or videos about their daily lives in social media, and many more (Aliu, Imran, Imran, & Evans, 2013).

In the future, however, this demand is expected to be much larger, with the advents of new technologies being developed, such as ultra-high definition videos, Virtual Reality (VR) applications, the Internet of Things (IoT); Machine-to-Machine (M2M) communications; cloud computing, and various other services that are unimaginable today. In addition, not only new technologies will need to be addressed by the Next Generation of Mobile Networks (NGMN), but also the ever increasing demand of users in terms of both capacity and better services (Huawei, Technologies Co., 2016), (Samsung Electronics Co., 2015). Some of the requirements that are present in the current state-of-the-art literature of NGMNs are (P. Fettweis, 2012), (G. Andrews, et al., 2014) (Huawei, Technologies Co., 2016):

  • Address the exponential growth required in both coverage and capacity;

  • Provide better Quality of Service and Experience (QoS and QoE) to end users;

  • Support the coexistence with other radio technologies;

  • Provide peak data rates higher than 10Gbps;

  • Support latency lower than one millisecond (enabling the concept of tactile internet);

  • Support ultra-high reliability and network energy efficiency.

From these requirements, it is clear that NGMNs are under a heavy pressure in order to address current limitations of present mobile networks and to push their performance to a next level, allowing all these requirements and new technologies to become a reality in the near future. In order to address the expected exponential growth required in both coverage and capacity, one major shift that will probably occur in NGMNs is the network densification process, in which operators are expected to deploy a wide range of small cells in order to cover a relatively small area (G. Andrews, et al., 2014), (Alsedairy, Qi, Imran, Imran, & Evans, 2015). Although the densification process is probably going to solve some of the future cellular network problems, this process will require an even bigger change in paradigm on how future cellular networks are organized. The dense deployment of small cells will result in an exponential increase in terms of network complexity, while also increasing the total network CAPital and OPerational EXpenditures (CAPEX and OPEX), affecting the whole network in terms of configuration, optimization and healing (Valente Klaine, Imran, Onireti, & Demo Souza, 2017).

Key Terms in this Chapter

Reinforcement Learning: Algorithms that learn from their own experience with the environment.

Self-Organizing Cellular Networks: Mobile networks that are capable of autonomously adapting itself to changes in their environment, while maintaining its desired objectives.

Self-Healing: Autonomous detection, classification, and management of failures in a mobile network, so that the effects of a fault can be mitigated.

Mobile Networks: Radio communication network, in which the last hop of the network is wireless.

Unsupervised Learning: A class of machine learning algorithms that, given a set of unlabeled data, is capable of learning a function that best represents the model.

Self-Configuration: Autonomous configuration of network parameters in order for a network to become operable.

Self-Optimization: The process of autonomously adjusting network parameters so that the network performance can be back to near optimum.

Supervised Learning: Class of machine learning algorithms that rely on the knowledge of an external supervisor in order to learn.

Machine Learning: Class of algorithms that analyze data and make predictions about it.

Complete Chapter List

Search this Book:
Reset