Advancements of QoE assessment and optimization in mobile networks in the machine era

Although QoE evaluation methodologies for human communication services in mobile networks - such as voice and video streaming - are a widely researched topic, several challenges remain yet unresolved including industry-agreed definitions of end-to-end performance indicators. With 5G, a wide variety of new use cases and diverse services shall be introduced, posing further challenges to the QoE assessment. For instance, emerging machine communication services may suffer from interdependencies across machines (i.e. systemic effects), which have yet to be fully understood. Further, to support reliable real-time services with ultra-high availability, future QoE optimization must take a more balanced approach across all network domains as opposed to the RAN focused approach of today. This paper discusses those challenges and presents the foreseen evolution of the QoE evaluation methodologies suitable to 5G mobile networks to accommodate the identified challenges. Building on today's best practices, we describe a flexible QoE service model and evaluation framework, which relies on Key Quality Indicators (KQIs) and allows for objective assessment of the QoE on a per service basis. At the same time, the model can effectively guide network optimization processes according to the network operator's strategy and actual network design. We validate the proposed framework in a live LTE network of a Tier-1 operator, demonstrating its capability to clearly identify the phases of performance optimization as well as we indicate how it could be applied to 5G to support machine-driven communications.