A Cloud-Based Architecture for Distributed Processing in Networked Games

  • Craig Hull

Student thesis: Doctoral Thesis


This thesis presents a framework for the improvement and maintenance of the QoS (Quality of Service) of networked video games. Hardware and software are monitored by certain variables and decisions are made based on the range of these. A high QoS is sought after by all players of video games, however delivering this has proven to be difficult at times. Cloud gaming technology has greatly improved distributed processing, though there are still factors inhibiting it. High load on servers and high round trip time to the user’s devices and consoles are preventing the users from achieving a high QoS. With games becoming more accessible, the range of devices that they can be executed on increases, though the quality varies from device to device. The servers which the game providers utilise can come under stress. For example, the hugely popular augmented reality game Pokémon Go came under fire from users as the servers could not handle the stress of the huge number of connected users resulting in server outages [1]. Another challenge is the issue of latency as many users may suffer from a low bandwidth internet connection which results in a poor user experience. It would be ideal for all game players to achieve a high QoS regardless of the device they are using, their connection to the server and the condition of the server. The research hypothesis underpinning the work described here is that cloud gaming techniques can be utilised to improve a user’s QoS.

A novel and adaptable architecture that combines cloud and fog assistance with self-adaptation techniques, in which the client adapts to a situation, is proposed as a solution to this problem. By employing available resources from the game server (cloud) and other under-utilised network nodes local to the device (fog), a game player’s QoS may be improved. Self-adaptation procedures are the last resort solution of the architecture should there be no available resources both locally and globally. Testing of this architecture is carried out under various conditions from varying latencies and packet loss to data packets of differing size being distributed and self-adaptation occurring due to different in-game elements. Results from experiments based on varying pressures in the game world and network conditions show that, by constantly monitoring the QoS of the game and the network, effective decisions can be made to improve a declining QoS. A smaller data packet transmitted frequently provides a greater improvement in comparison to a larger data packet transmitted less frequently.
Date of AwardMay 2018
Original languageEnglish
SupervisorDarryl Charles (Supervisor), PJ Morrow (Supervisor) & G Parr (Supervisor)

Cite this