A framework for adaptive residual streaming for single-player cloud gaming

Abstract

Applying cloud technology to 3D interactive multimedia applications is a promising way to provide flexible and cost-efficient online high-bandwidth immersive services to a large population of end users. One main reason cloud systems are popular among users is the fact that it relaxes the hardware requirements for highend interactive visual applications. As most of the computational tasks are done on cloud servers, users no longer need to upgrade their hardware as frequently to keep up with the ever-increasing high-end computing requirements of the latest applications. Moreover, cloud systems make it easier for a user to enjoy applications on different platforms, including mobile devices that are usually not powerful enough to run high-end, memory-intensive services. In short, applying cloud technology to high-end immersive applications has advantages in cost-efficiency and flexibility both for the end users and the service providers. However, there are two main drawbacks to applying cloud technology to 3D interactive multimedia services: (1) high-bandwidth utilization and (2) latency. In this article, we propose a framework that addresses the two problems for singleplayer cloud gaming by using a combination of collaborative rendering, progressive meshes, and 3D image warping techniques. The experimental results show that the proposed system can reduce the bandwidth usage and improve the visual quality by utilizing local computing power on the client. The results also show that the interaction latency can be reduced somewhat by sacrificing some degree of visual quality in the end system.