Fujitsu has developed a technology that allows for virtual desktops with larger screens, without upgrading existing network infrastructure. Virtual desktop technology runs the user's applications on a virtual server in a cloud environment and transfers desktop images to thin clients or other terminal devices. With a virtual desktop, bigger desktops mean more network bandwidth is used when sending desktop images. In order to increase virtual desktop sizes without incurring the higher cost of enhancing network infrastructure, virtual desktop images need to be transmitted more efficiently.

Fujitsu Laboratories has applied the high-quality video coding technology cultivated by Fujitsu to virtual desktop screen compression, and has developed technology that maintains image quality while cutting network bandwidth requirements roughly in half. The company says that the new technology makes it possible to transmit roughly double the pixel count over existing network infrastructure, allowing for bigger, more detailed virtual desktops, and a more functional work environment.

This technology is due to be included in FTCP Remote Desktop, an engineering cloud, to be released in May 2015.

Technology details

The company developed an optimization technology that compresses data while maintaining the high quality suited to the characteristics of virtual desktop screens, and that reduces processing load in response to image content.

Unlike broadcast images shot with a camera, a virtual desktop screen, with its extremely sharp lines drawn with CAD, CAE or other applications, has the characteristic of completely still-image backgrounds or other spaces. The technology maintains linear sharpness by optimized encoding mode selection, including block size and intra prediction mode, and largely obviates the processing load for still-image regions by optimized motion vector searches.

Typically, static areas and moving areas within an image to be transmitted would be flagged separately, and appropriate compression schemes would be applied to each to maximize the overall compression of the data to send. This approach, however, results in increasing the volume of data when static and moving images frequently switch. To solve this problem, Fujitsu developed a function that shares encoded images for reference between still-image coding and video coding* This has the effect of reducing data volumes needed for switching, even when regions of the image to send frequently switch between still image and video, as is the case with screens where CAD or CAE software is being manipulated.