Quality of Experience Model for Addis Ababa Voice Service Using Adaptive Neuro Fuzzy Inference Approach

This thesis deals with the development of a Comprehensive Quality of Experience (CQoE) which is the combination of objective Quality of Experience (OQoE) and the Subjective Perception of Quality (SPQ) in the ratio of 80% and 20% respectively. QoE (Quality of experience) is the perceived quality of a service by the user. CQoE is a model that measures and predicts the users' perceived quality. It uses the real Global System for Mobile Communication (GSM) and Universal Mobile Telecommunications System (UMTS) data, for this purpose, to channel it according to the need of the user. The ethio telecom has to be needed a model for QoE that is capable of assessing, measuring, predicting and anticipating the needs and expectations of customers. In a competitive environment, a service provider (SP) with no QoE model would easily lose its business to the others who have and use QoE models and strive sustainably to enhance the quality of the voice call service they provide. The user perceived quality is measured by Mean Opinion Score (MOS) scale, by evaluating Quality of Service (QoS) metrics. Service monitoring tool data and calculated MOS measurements are then used as reference data, in order to produce a new QoE prediction model, using Adaptive Neuro Fuzzy Inference System (ANFIS). The concept of the SPQ has been introduced with the intent of getting first hand feedback from customers, by using Questionnaire. Voice call Service is given to customers, who are human beings having two main aspects: the rational, objective and the emotional, subjective. OQoE deals with the rational, objective aspect of voice call service quality given by a SP, while the SPQ deals with the undeniable emotional, subjective aspect of voice service quality given by a SP. By combining these two aspects, one gets the full picture of voice quality level given by a SP. The 3G and 2G voice calls QoE models perform the perceived quality by these QoS metrics: PCSR ( Perceived call success rate), E2ED (End to End delay) and PCDR(Perceived call drop rate). These models estimate the QoE with a Root Mean Squared Error (RMSE) and Pearson Correlations in the following order respectively: for OQoE (3G & 2G), CQoE (3G& 2G) : 0.47% & 95.05%, 0.35% & 93.82%, 0.06% & 97.88%, 0.023% & 99.24%.to the measured QoE.