scope:

This Recommendation describes a model that delivers predicted mean opinion scores (MOS) on a 5-point absolute category rating (ACR) scale, see [ITU-T P.800.1], [ITU-T P.910], based on the impact of impairments introduced by typical Internet protocol (IP) networks on the quality experienced by players using a cloud gaming system. This Recommendation targets cloud gaming services that perform video streaming over real-time transport protocol (RTP) (over user datagram protocol (UDP)) and which select various video encoding parameters to adapt to the network throughput, packet loss, and end-to-end delay.

The focus of the described model is to predict gaming quality of experience (QoE) by considering relevant factors that are identified and discussed in [ITU-T G.1032]. The impairment factors are derived based on network and encoding parameters. By analysing the suitability of a variety of quality features for the prediction of the overall gaming QoE, an impairment model inspired by the E-model [b-ITU-T G.107] was developed.

The model is a network planning tool which can be used by various stakeholders for purposes such as resource allocation and configuration of IP-network transmission settings such as the selection of resolution and bitrates, under the assumption that the network is prone to packet loss and latency.

The model offers two different modes: a default mode where no information about the game type is considered, and an extended mode, for which various impairment factors based on a content classification with respect to the encoding complexity as well as the delay and frame loss sensitivity of a game are considered. Depending on whether the respective stakeholder (cloud gaming service provider or the network planner) has a priori knowledge of the type of game being offered through the cloud gaming service, the appropriate mode can be used. More information on the modes is given in clause 6.2 and Annex A.

Virtual reality games requiring 3D rendering devices, mobile input, and output devices, as well as input devices other than keyboard and mouse are not within the scope of this model. Nevertheless, the model might apply to such systems as well. This is an item for further study. The model is also not designed to predict the influence of the game design or the motivation of users to play them. The subjective ratings collected to develop the model are primarily derived from non-expert gamers and hence, the model predictions might not be accurate for highly experienced gamers due to their different expectations and sensitivity towards the various degradations. Furthermore, the influence of social factors are not considered in this model. While the focus of the described model is on cloud gaming, some parts may also be relevant for online gaming (where the game is primarily executed on the client) or passive gaming video streaming (where only the video content is streamed to passive viewers of the game) applications. With respect to the technologies considered, the model addresses cloud gaming services using graphics processing unit (GPU) hardware accelerator engines for video compression, and H.264 [ITU-T H.264] as the video compression standard.