Improving bandwidth and time consumption in remote visualization scenarios through approximated Diff-Map calculation

摘要

In the recent years, new business models of provision of services are grown for different kind of software thanks to the development of cloud environments. A large number of companies use 3D software tools and can be interested in adopting 3D visualization as a functionality of the cloud infrastructure rather than the single employee's workstation, saving hardware resources and licenses costs. Remote visualization solutions allows to decouple computing resources on one side and visualization resources on the other side; rendering servers are in charge to compute and stream to the user's device the output graphics of the intended software as a sequence of individual frames. However, the provision of highly graphics-oriented software within a cloud can be particularly challenging due to a number of technological limitations from bandwidth requirements to latency-related issues. In this scenario, the calculation of the differences between successive frames (Diff-Map) is of paramount importance to reduce the bandwidth usage and computational complexity. Diff-Map calculation is time and resource consuming, therefore a method for the calculation of an approximated Diff-Map is proposed. This paper, after analyzing the main recent technological trends followed in the research domain for the development of effective remote visualization solutions, addresses the analysis of network bandwidth usage and computational speed by adopting approximated Diff-Map methods for encoding and transmitting only relevant data pertaining to changed areas of the frame buffer. A set of experimental tests demonstrates the savings in bandwidth consumption and performance speed up going towards a quality tradeoff, alleviating the burden of transmitting excessive useless data to remote client viewers.