Predicting QoE for Delay-critical Services in Mobile Networks: A Video Conferencing Case Study.

There is an increasing demand to support high-quality data communications for delay-critical services such as Augmented/Virtual Reality (ARNR), cloud gaming, drone control or video conferencing. Mobile network operators need to estimate quality of experience (QoE) values based on available QoS parameters such as radio environment and core network metrics, to respond to performance degradations and maximize revenue, as the QoE of users is more important than the QoS metrics. To develop accurate QoE models, operators need a large amount of versatile data under different network conditions. We propose a standardized and automated data collection framework to gather fine-grained transport and radio data, along with user experience values assigned through a standardized survey, for model training purposes in delay-critical services. Our study identifies valuable QoS parameters and relationships between mobile network degradations and service quality impacts, demonstrating the possibility of building accurate QoE models with an R 2 of 0.78 based solely on transport and radio data. Our finding can be beneficial to MNOs in designing new data collection and QoE estimation services for various delay-critical services.