QoE Model for Social Media Video Streaming Service Using Ensemble Method - The Case of Addis Ababa

No Thumbnail Available

Date

2022-01

Journal Title

Journal ISSN

Volume Title

Publisher

Addis Ababa University

Abstract

Nowadays, the number of Social Media (SM) users is increasing tremendously worldwide. An increase in the number of smartphone users and an increase in Internet coverage helps people expand their networks. The availability of SM helps users to find their friends, make a connections with new people with different skills, and improve their careers. The diversity of services on SM also attracts many new users to use SM in their day-to-day activities. To deliver SM services, many SM application developers continuously work to improve their services and also to add new services to attract new customers and also to keep their users. There are many stakeholders that are involved in end-to-end service delivery of SM. These include Telecom network providers, SM application owners, and end-user devices performance. Having a good network Quality of Service (QoS) may not guarantee good service quality on the customers’ side. The quality of a given service perceived by an end-user, which is Quality of Experience (QoE) is a broad term and influenced by many Influencing Factors (IF). There are many research papers done on the QoE model of different SMs. Most of the papers focus on the impact of network and application related QoS on the overall customer satisfaction level. Even though these papers incorporate different parameters as input features for the QoE model, to the best of the author’s investigation, researches which are done in the context of Ethiopia didn’t consider the users device parameters influence on the customers’ satisfaction level. The over all QoE of a service is influenced by many factors, the main focus of this thesis is to provide a QoE model for SM video streaming services by taking different IF as input parameters. From the network QoS parameter download speed, upload speed, latency, and jitter are used as inputs. From users’ device parameters phone Random Access Memory (RAM) size, phone’s free internal storage size, and phone’s screen resolution are taken as device IF. And from application parameters, video resolution is used as an input for the model. The developed model is based on an ensemble technique which is a Machine Learning (ML) based approach. The model has good accuracy, which is 94.1% accuracy. In addition to the accuracy, based on the importance of each input feature to the final model, download speed takes the main influencing share by 52.357% from the total input parameters and from the users’ mobile device parameter free internal storage space has 10.784% and mobile RAM size 9.4% on the final QoE model. Generally, this work meets the initial objective by developing QoE model with a good accuracy, and shows the influence of other parameter other than the usual network QoS parameters and gives insight to the gaps that cause the customers’ dissatisfaction of the SM video services by considering different IF.

Description

Keywords

QoE, QoS, Social Media, Ensemble Method, Machine Learning

Citation