QoE Issues of OTT Services over 5G Network

Abstract

Currently ， 4G network becomes commercial in large scale around the world and the industry has started the fifth-generation mobile communication technology (5G) research. Compared to 4G network, 5G network will support larger mobility as well as higher transmission rate, higher user experience rate, energy efficiency, spectrum efficiency and so forth. All of these will boost a variety of multimedia services, especially for Over-The-Top (OTT) services. So for, OTT services have already gained great popularity and contributed to large traffic consumption, which propose a challenge for operators. As OTT services are designed to deliver over the best effort Internet, the QoE management solutions for traditional multimedia services are obsolete, which propose new challenges in QOE management aspects for network and service providers, especially for the 4G and future 5G network. This paper attempts to present the technical challenges faced by 5G network from QoE management perspective of OTT services. Our objective is to enhance the user experience of OTT services and improve network efficiency. We analysis the characteristics and QoE factors of OTT services over 5G wireless network. With the QoE factors and current QoE management situation, we summarize OTT services QoE quantification and evaluation methods, present QoE-driven radio resource management and optimization solutions. Then, we propose a framework and whole evaluation procedure which aim at obtaining the accurate user experience value as well as improving network efficiency and optimizing the user experience.

Challenges of future multimedia QoE monitoring for internet service providers

“…This category is referred to as "context-aware", since the (user) context has become essential in the modeling of the user-perceived quality. Algorithms based on machine learning and artificial intelligence are gaining interest in this scenario due to their potential for higher adaptability, reliability and robustness over statistical and parametric/probabilistic models [15].…”