Edge computing is perfect for enabling breakthrough applications on power and compute constrained devices like today’s emerging and rapidly evolving VR / AR / XR headsets.
Consumer VR headsets have advanced to the point where everyone expects market volumes to grow rapidly. But these consumer devices are optimized for consumer price points and often don’t have the adequate speed or memory capacity to support valuable commercial applications that are emerging in parallel. For example, Microsoft’s HoloLens was just cleared for surgery use by the US FDA; this mainstream acceptance into the mainstream shows the maturity and utility of the technology. But the capacity of the consumer device just isn’t quite there, and the more powerful headsets don’t benefit from consumer volumes and are much more expensive. For example, if you try to transmit a large 3D CAD model to HoloLens, the limited memory of 1GB is too small given application setup takes about 200MB, leaving the truly available memory at only 800MB.
Wouldn’t it be nice if there were some way to have your cake and eat it too? Is there some way these economical, rapidly advancing consumer devices could be used for more complex commercial applications as well? With edge computing they can!
Holo-Light is a company that has created specialized software for the visualization of 3D content that leverages edge computing to deliver a significantly better visualization user experience than consumer devices can by themselves
Edge Compute to the Rescue
Mobile device design is complex, full of conflicts, and unforgiving. You want a mobile device to have compelling functionality (display quality, processing power, large memory) and you want the device to be attractive and as light as possible. The biggest single problem is the battery. Greater functionality requires greater power; battery life is determined by power output and usage time. Bigger batteries are proportionally heavier. You can’t have it all; you have to make tradeoffs between capabilities, weight and battery life.
That’s where edge computing comes in. Unlike the computers in devices, computers at the edge don’t have to be battery powered, and the weight and size doesn’t matter, so we can offload computation tasks from the device to the edge and have the best of both worlds. How well this works depends on the latency and bandwidth of the connection between the device and the edge computer. As wireless communications technology improves the bandwidth goes up. As the compute is moved to the edge, minimizing the networking in between diminishes, latency goes down and bandwidth goes up, enabling all sorts of exciting new application architectures. Edge computing in mobile networks shows how this enablement is possible.
By offloading the most demanding computation to the “edge”, we no longer have to tradeoff capabilities against headset weight -- we can do those tasks where power is free and send them to the device within the latency constraints of realistic AR and VR.
The MR/AR device sends positioning and orientation data to the edge server and transfers the results into its graphics subsystem. If this happens fast enough - around 30 - 60 frames per second (fps) - the result is indistinguishable from the application running on the device itself. The free power at the edge gives us the option of putting more graphics capability than is possible in the device, in which case, the visualized content appears to be even more realistic.
By rendering 3D content remotely on this edge computing infrastructure with low-latency, high-bandwidth interconnection, Holo-Light is able to visualize massive models on Microsoft’s HoloLens as was demonstrated at Hubraum events in Krakow and Berlin and most recently Bonn in June. The end result went well beyond what could be done on the device, even if the storage limitation were fixed -- it was visibly a better solution.
The Edge Computing Solution
MobiledgeX is bringing developer-friendly, cloud-like service capabilities to the edge of the mobile network, and that in turn enables companies like Holo-Light to build innovative new applications that integrate edge computing. MobiledgeX has already installed an early version of the platform in the DT infrastructure in Germany and plans to add more mobile operators and regional coverage rapidly in 2019, thereby, in essence, pushing cloud computing much closer to the edge and creating an ever-growing footprint for this kind of innovative application.
Check out our previous blog posts highlighting edge use cases: