Mapping QoS to QoE Using Hidden Markov Models: A Case Study on Virtual Internet Service Provider Networks
No Thumbnail Available
Date
2025-06
Authors
Journal Title
Journal ISSN
Volume Title
Publisher
Addis Ababa University
Abstract
For Internet Service Providers (ISPs) and virtual ISPs (vISPs), accurately understanding and measuring Quality of Experience (QoE) is essential, yet doing so is a challenging task because they have relied solely on technical Quality of Service (QoS), which fails to capture the true QoE perceived by users due to the user-centric, individual, multidimensional, and multisensorial nature of QoE. Therefore, this has led to the evolution of existing predictive models that map QoS to QoE. However, they are often limited in capturing sequential patterns or hidden transitions in user-perceived quality over time. This thesis proposes a Hidden Markov Model (HMM)-based approach to model the mapping between QoS parameters and QoE metrics using objective data and a composite of the two (objective and subjective data) for both ethio telecom and vISPs (Websprix IT Solution PLC and ZERGAW Cloud). The Baum-Welch algorithm was used to train the model, and its performance is compared against Support Vector Machines (SVM) and Random Forest (RF) models. The results of the study convincingly demonstrate the performance scores of the HMM prediction model with 99% RF with 96%, and SVM with 71% accuracy, which outperforms the SVM and RF models. Also, HMM had around 100% in all precision, recall, and F1- score, particularly in scenarios with high network variability characteristic of vISP networks. These findings highlight the potential of HMMs for improving QoE prediction in service
provider networks and supporting more user-centered service optimization strategies.
Description
Keywords
QoS, QoE, Hidden Markov Model, Support Vector Machine, Random Forest, ISP, vISPs