BasePaper SQM
BasePaper SQM
BasePaper SQM
№ 682
Abstract—Network Operators are shifting their business improve service quality taking into consideration the time
interest towards Data services in a geometric progression manner, constraint. The justification of the big money being invested in
as Data services is becoming the major source of Telco revenue. Service Quality and Customer Experience Managements is
The wide use of Data platforms; such as WhatsApp, Skype, scaled on multiple levels including Domain expertise,
Hangout and other Over the Top (OTT) voice applications over technology awareness and the investment in Science,
the traditional voice services is a clear indication that Network Engineering, Mathematics and Technology. In this paper, a
Operators need to adjust their business model and needs. And model of a Tree-based Service Quality Management system is
couple with the adoption of Smartphones usage which grows investigated and illustrated using data collected from an
continuously year by year, which means more subscribers to
Operator, SparkSQL with In-memory Big Data processing and
manage, large amount of transactions generated, more network
resources to be added and evidently more human technical
cloud based BI tool for visualization.
expertise required to ensure good service quality. With the large
amount of transactions generated by data traffic and the high II. THE CONCEPT OF IN-MEMORY BIG DATA
demanding service qualities, Mobile Network Operators are
The concept of Big Data is related to the amount of data to be
spending millions of rands/dollars to deploy Robust Service
Quality Management (SQM) and Customer Experience processed. The processing of such data requires a lot
Management (CEM) to stay competitive in the market. These computational power and advanced technologies. The
solutions’ high cost is justified by the integration of Big Data characteristics of Big Data emphasizes on the Volume, Velocity
Solutions, Machine Learning capabilities and good visualization of and Variety “VVV” techniques, called the three “Vs” to make
insight data. However, the Return on Investment (ROI) of the the difference with the traditional data handling methodologies,
expensive systems are not as conspicuous as the provided Volume, Velocity and Variety [2].
functionalities and business rules. Therefore, in this paper an
efficient model for low cost Service Quality Management system is
presented, using the advantages of the In-Memory Big Data Amount of Data to be
Data Volume retrieved
processing and simple low cost business Intelligence tools to
showcase how a good Service Quality Management can be
implemented with no big investment.
Data Velocity Data Retrieval Speed
Keywords—Service Quality Management, In-Memory Big Data,
Business Intelligence, Service Quality Index, Over The Top
Data diversity (Files, images,
Application (OTT), Data Traffic and ROI.
Data Variety videos …)
I. INTRODUCTION
Figure 1: Big Data Characteristics, the 3 Vs
One of the biggest concerns of the Cellular Network
Operators is to ensure continuous service availability and quality
for a good subscriber retention index. However, with the growth A. In-Memory Data Processing
and high adoption of Smart phones and Smart devices alike, For many years data processing has been based on data
increase in Data services, data traffic is increasing exponentially. stored in disks including relational databases such as Structured
Tracking Service Quality becomes a tedious and complicated Query Language (SQL). However, with the growth of data and
exercise, requiring not only telecommunications knowledge but the high demand of low-latency services, the concept of in-
also the knowledge required to find useful information in the memory Database storage has developed. Because data stored
billions of transactions records generated by mobile subscribers. in the memory are accessed quickly, in-memory processing
Thus, the involvement of Data Analytics, Big Data processing becomes the most suitable methodology for real time analytics
and Business Intelligence in the area of Telecommunications.
[3]. In-Memory data processing has the advantage of reducing
The big step in terms of innovation in the area of technology is
the Hardware footprint while improving the performance. CPU
the advancement of computer technology which has changed
completely the way of handling Data, Analyzing Data and power, Memory and disk storage are considered and are subject
visualizing data [1]. The data collected from the network to variation to meet the business need.
contains a lot of subscribers’ information that can help network
operators in decision-making processes and business
improvement methodology. Cellular Network Operators aim to
B. SparkSQL Another important aspect of BI is the drilling down and
The advance of technology comes with the challenge of navigation between different dashboards. In this paper, the
skills development. With the high success in the market of concept of Service Quality Index (SQI) is introduced along the
traditional relational database systems (RDBMS), companies mentioned key aspects.
have invested a lot in relational databases such as SQL.
Migrating completely away from the Relational Databases to Measurement Result
KPI
new architectures becomes a huge technological challenge.
SparkSQL is an Apache Spark module developed to process
Relational Data Structure with Spark [4]. It allows the usage of Aggregation of KPIs
SQL queries to retrieve information. Thus, SQL Users to take Scorecards
advantage of the Big Data processing. SparkSQL presents three
main capabilities [5]: Information Overview
Data frame abstraction in different programing Dashboards Window
languages including Java, Python and Scala to
efficiently work with structured data sets.
Drilldown
Read and write many popular structured data formats
such as JASON, Parquet, and Tables).
Using SQL, query data from inside Spark program and
from external tools that use database connectors to Figure 3: Key Concepts of Business Intelligence
communicate with Spark (JDBC/ODBC). This scenario
is used in this paper since the connection to the dataset IV. PROBLEMATIC AND OBJECTIVES
is via JDBC connector.
Using the advantages of SparkSQL with in-memory data
caching and the BI tool, this paper demonstrates the building of
a low cost tree based Service Quality Management for Network
MySQL, Access, evaluation and monitoring. The objective is to construct a basic
Oracle SQL framework that leverages on the current business demand of
Network Operator. The business needs of Operators differ
Spark Shell based on the business models, the processes, competitors and
JDBC/ODBC
maturity in the business. Nevertheless, the main goal and
objective is to have good Service Quality and improve
Customer Experience.
SparkSQL
Table 1: QoS Service Class VIII. UNDESRTANDING THE DATASET AND KPI
The dataset contains service information from the network. It is
Service Characteristics Applications made of categorical fields that are used as keys for queried
Class information and numerical fields from which KPIs and Score
Conversatio Low delay. Voice Call cards are deduced. The data fields used in the SQM database
nal Preserve time VoIP system contains user information, location information,
variation application information and Key Performance Indicators.
Streaming Preserve time Audio & Video
variation Streaming A. Key Performance Indicators
Interactive Request response Web browsing
The exchange of information in the network is done through
pattern Instant transport and application protocols which carry packet. TCP
Preserve payload Messaging (Transmission Control Protocol) is used as the transport
content protocol for connection-oriented packets. TCP applications are
Background Not delay sensitive Email very sensitive to the conditions of the communication pipes.
Preserve payload Files The selected performance metrics used for the SQM are as
content download follow:
Processing Server
BI Tool
In Memory
Cache
Customer SparkSQL
DB
Disk Storage