A Neural Network for Quality of Experience Estimation in Mobile Communications

Mobile communication systems increase data rates for the growing number of subscribers using ubiquitous data services. To provide better satisfaction to the users, high throughputs have been usually envisaged by network operators. In recent years, due to the growing number of tablets and smartphones, the push applications and connections to social sites, the user needs require low throughput. A new analysis of user satisfaction is necessary, the so called Quality of Experience (QoE). Recently, new methods have been proposed to estimate the QoE through the specific key performance indicators (KPI). The paper considers various KPIs built on data acquired from the high speed downlink packet access network and analyzes their correlation with respect to the QoE. The use of neural networks is proposed to provide an automatic classification among these KPIs (related to Quality of Service) and QoE for mobile internet services. The performance of different neural networks are compared and then the related QoE estimation is shown in terms of Mean Opinion Score. The adoption of the neural network ensures replicability of QoE estimation regardless of user involvement and an easy classification of user satisfaction for the future fifth generation systems, only through a new training data step.