Metaheuristic-based Workload Selection for Hybrid Cloud Rendering of CAD Models

EasyChair Preprint 1700

Abstract

Efficient rendering of massive industrial-plant CAD models is a long-time challenge even when using high-end computers. This problem has been aggravated with the popularization of web-based applications and lightweight devices (e.g., tablets and smartphones) in engineering daily basis activities. One promising solution consists in pushing rendering tasks to the cloud, similar to recent cloud gaming approaches. However, existing cloud gaming platforms are not able to render CAD models efficiently. These platforms are mostly concerned about game synchronization and latency reduction, not supporting massive models. In this work, we aim to fill the gap between industrial-plant models and cloud rendering by proposing a hybrid cloud rendering method. In our approach, we define two sets of objects: one for rendering on client-side, and the other for rendering on server-side. The client produces the final image combining local and remote images. As a consequence, our cloud rendering system provides better image quality and more robustness to network fluctuations. A key component of such a solution is the choice of objects to be rendered on the client. We propose a metaheuristic-based method to select such objects. Our solution privileges larger objects with a uniform spatial distribution. This way, the user can still coherently navigate even when the server is not responding. Our results show that our approach enables thin devices to interactively visualize massive CAD models with good image quality, even when the networking is not performing well.

Keyphrases: CAD models, Hybrid Cloud Rendering, Workload Selection, cloud rendering, industrial plant cad model

@booklet{EasyChair:1700,
  author    = {André Moreira and Waldemar Celes},
  title     = {Metaheuristic-based Workload Selection for Hybrid Cloud Rendering of CAD Models},
  howpublished = {EasyChair Preprint 1700},
  year      = {EasyChair, 2019}}