Measuring and estimating Key Quality Indicators in Cloud Gaming services

Abstract

The gaming industry has proposed the concept of Cloud Gaming (CG), a paradigm that enhances the gaming experience on reduced hardware devices. However, this paradigm puts a lot of pressure on the communication links that connect the user to the cloud. As a result, the service experience becomes highly dependent on network connectivity. In this context, the present work proposes a framework for measuring and estimating the most important E2E (end-to-end) metrics of the CG service, namely Key Quality Indicators (KQIs). Therefore, different machine learning (ML) techniques are evaluated to predict KQIs related to the CG user experience. For this purpose, the most important KQIs of the service, such as input lag, freezes or perceived video frame rate, are collected in a real network deployment. The results show that ML techniques can be used to estimate these indicators solely from network-related metrics. This is seen as a valuable asset for the delivery of CG services over cellular networks, even without access to the user’s device, as it is expected for telecom operators.