A foveal vision based spatial prefetch method on digital earth platform for performance improvement

Higher performance of digital earth system is still demanded in order to always guarantee high quality service especially while large-scale network users request intensive data such as 3d city model with fine texture in large scene. This paper aims at extracting information from user´s behavior to guide prefetch which can help enhance service capacity of digital earth platform. We put forward three aspects work to effectively support and make sure pre-fetch useful. They are effective pre-fetch mode, peer-to-peer (P2P) environment support and priority-based multichannel messaging technology. Pre-fetch mode defines priority allocation model of data request which is inferred from the current view of request data. We designed simple pre-fetch mode and foveal pre-fetch mode, and testify pre-fetch methods in many cases to show effectiveness of performance improvement on an abstract P2P simulation system with setting different network speed, random user exploring path and different exploring moving speed.