Prediction of video quality degradation on a cloud gaming platform.

Abstract

Cloud gaming has become a promising way to play high-quality video games on relatively inexpensive devices. Cloud gaming platforms run interactive video games remotely in the cloud and stream multimedia to the gamer’s device via a telecommunications network. These platforms have many benefits, but also pose several problems and challenges, as they are highly dependent on the network from servers to users. This work aims to study the effect of network impairments on video quality, and to predict video quality degradation in real time. In particular, it was determined that latency is a good predictor of packet loss, which was found to result in visual degradation. Based on this, an algorithm was designed to predict real-time visual degradations before they actually occur.