Prediction-based Prefetching for Remote Rendering Streaming in Mobile Virtual Environments

Remote Image-based rendering (IBR) is an effective solution for rendering complex 3D scenes on mobile devices. A server renders the 3D scene and streams the rendered images to the client. However, sending a large number of images is inefficient due to the possible bandwidth limitations of wireless connections. In this paper, we propose a server-side prefetching scheme that predicts client movements and hence prefetches the corresponding images. In addition, an event driven simulator was designed and implemented to evaluate the performance of the proposed scheme. The simulator was used to compare between prediction-based prefetching and prefetching images based on spatial locality. Several experiments were conducted to study the performance using different movement patterns and different virtual environments (VEs). The results have shown that the hit ratio of the prediction-based scheme is greater than the localization scheme in the case of random and circular walk movement patterns by approximately 35% and 17%, respectively. For a VE with high level of details, the proposed scheme outperforms the localization scheme by approximately 13%. However, for a VE with low level of details, the localization based scheme outperforms the proposed scheme by only 5%.