A Multiplayer Cloud Gaming Architecture for Scalable Physics

Utilizing cloud computing resources, cloud gaming, or Games-as-a-Service (GaaS), has become a novel game delivery approach, offering gaming accessibility across various devices anytime. It involves shifting computation-intensive tasks to the cloud, addressing hardware limitations. However, due to a monolithic design, many proposed cloud gaming architectures face challenges in fully exploiting cloud computing resources for scalable systems. Additionally, there is a lack of consideration for how multiplayer games could harness the potential advantages of this innovative paradigm. This paper proposes a multiplayer cloud gaming architecture that leverages a service-oriented design to provide scalable physics. As a contribution, our work considers multiplayer cloud games while also provisioning the physics engine as a decoupled service that can be scaled according to the game’s physics workload. By implementing a proof-of-concept, we could observe that the service-oriented approach resulted in better game server performance, in which one experiment produced approximately $68 \%$ lower total physics processing time. This happens because the more physics services added, the more significant the performance gains were. This indicates the benefits of decoupling the game engine physics system, scaling better both vertically and horizontally.