Just as the consumption of series or films on streaming platforms has become widespread, people are increasingly turning to online video games, or cloud gaming. But, unlike other audiovisual products, online videogames are interactive, posing added technical difficulties to the platforms that supply them on the Internet to optimize their functioning via the web.

Faced with this challenge, a study by two UPF researchers published in a scientific article recently granted the Best Paper Award 2022 by the journal Computer Communications, offers an in-dept examination of the workings of a cloud gaming application. The research gives a detailed analysis of how the network protocols used by the application are able to adapt to Internet’s varying capabilities.

The title of the article is “Cloud-gaming: Analysis of Google Stadia traffic”, by Boris Bellalta, director of the Wireless Networking (WN) research group at the UPF Department of Information and Communication Technologies (DTIC), and Marc Carrascosa, a researcher with the WN. The research looks at the case of Stadia, the Google cloud gaming platform, whose data traffic generation patterns are examined in depth.

Based on Stadia, the study analyses how cloud gaming platforms generate traffic (the bits per second sent to the network for each moment in time) of video and audio data and how they adapt to the available Internet capacity at any given moment. Since Internet is shared by many users and different applications, the capacity of its communication links greatly fluctuates over time. Hence, cloud gaming platforms must adapt the traffic generated to these changes as quickly as possible if they wish to provide a good gaming experience to the end user.

Marc Carrascosa, main author of the article and WN researcher, considers that research in this field is essential to improve the quality of cloud games in the future: “Cloud gaming applications have far more restrictive network performance requirements than other current video applications due to the real-time interaction with users. Understanding and modelling their operation will enable us to optimize the networks of the future to provide better service to users”.

To optimize the traffic of data of cloud gaming platforms, artificial intelligence techniques are also being used. So explains the principal investigator of the research and director of the WN, Boris Bellalta: “We are currently investigating the use of mechanisms that use artificial intelligence and machine learning techniques to improve the performance of today’s systems, with the end goal of offering a better experience to users”.

For Bellalta, this line of research also aims to improve the quality of a service that facilitates access to video games, which is possible from any device with an Internet connection and does not require specific terminals like a console. “The online gaming model is associated with moving all computing capacity to the cloud, which makes user terminals far simpler and more economical. However, there is a need to manage how data is exchanged between the cloud and users, and design mechanisms capable of maximizing the use of available bandwidth in a situation of great uncertainty regarding how it will fluctuate in the future. It is in this context that the use of machine learning techniques can gain prominence, thanks to its ability to make accurate predictions based on the accumulated experience”, he explains.

The study shows that Stadia adapts appropriately to changes in Internet’s ability to transmit data. However, there is room for improvement with regard to Stadia’s recovery time in the face of sudden drops in network capacity.

The research also concludes that Stadia’s data traffic pattern is always the same regardless of the features of the video game with which it is interacting at any given time. Each video game can generate a different volume of data (more or less bits per second), but the rate at which information is transmitted (time between packets) over the Internet always follows the same pattern.

The analysis of Stadia has provided the researchers with valuable information to plan and size data traffic generation patterns for cloud gaming platforms in general. Based on this study, they have defined a data traffic model to replicate the operation of Stadia, which could be generalized to other cloud-gaming platforms and be useful for other research.

As a challenge for the future, the UPF researchers plan to examine and improve the operation of cloud gaming platforms like Stadia when accessed through Wi-Fi networks. The specificities (such as the strength of the Internet connection signal, the volume of users connected at the same time, the presence or absence of other nearby Wi-Fi networks...) require examining in greater depth to optimize the use of cloud gaming in wireless networks. With this specific goal, the WN group is participating in the MAX-R and WI-XR projects, funded by the European Commission and the Spanish Government, respectively, which also study the virtual reality component.