QoE Management made uneasy: The case of Cloud Gaming

In this paper we discuss challenges and requirements for QoE Management in future networks using Cloud Gaming as representative paradigm for upcoming services. In particular, we highlight inherent difficulties for QoE Management such as an extreme variation in terms of QoE sensitivity and resource requirements within a single application category. In this context, motivated by contradicting results in related Cloud Gaming QoE work, we analyze the relationships between delay, latency detection and QoE and discuss possible influences on user delay-sensitivity. We also present the results of a laboratory cloud gaming QoE study showing that even small additional delays can have a visible QoE impact while on the other hand users can be surprisingly tolerant to higher delays. Indeed, the observed QoE sensitivity strongly depends on which game is actually used, representing a typical characteristic of cloud gaming and a key challenge for QoE Management. Based on our results, we discuss how QoE Management needs to take the identified diversity of user- and application-related factors into account, reaching far beyond just satisfying static network QoS demands. Using the exemplary case of latency management, we argue that QoE Management for emerging services like Cloud Gaming requires a holistic end-to-end approach as well as a structured information exchange between applications and networks, both essential features of Future Internet infrastructures.