At Mobile World Congress this morning, Deutsche Telecom, SK Telecom and Ericsson showed a telepresence-controlled robot prototype to demonstrate the capabilities of 5G wireless networks, software-defined networks (SDN) and Network Functions Virtualization (NFV).
The point made is 5G’s ultra-low latency delivered end to end with SDN and NFV will enable new real-time use cases for industrial services, autonomous cars and the Internet of Things (IoT). Within a 5G radio access network (RAN), 5G can deliver sub-1ms latency. But a loaded RAN or traversing a network between RANs can add tens of milliseconds—too much for control applications that require synchronization over distances.
Deutsche Telecom, SK Telecom and Ericsson demonstrated the telepresence and control prototype created by Frank Fitzek, head of the 5G lab at the Technical University of Dresden, and his team. The demonstrator wore a jacket and glove with several built-in sensors that mapped his movements onto a robot’s arm and an Oculus virtual reality (VR) headset that mapped his head movements onto the robot’s head and displayed the robot’s cameras field of view, including 3D depth perception. The robot copied the demonstrator’s movement with tight synchronization. The video below explains the choreographed movements of the demonstrator and the robot.
Video Credit: Deutsche Telekom
Two cases were demonstrated. In one case the robot was controlled by the demonstrator over the wireless network managed with the SDN and NFV and the other without the SDN and NFV. In the first case, the demonstrator was able to synchronize the robot’s movements closely to his. In the second case, where the SDN and NFV were unavailable and caused increased latency, the demonstrator could not coordinate the robot's movements with his.
SDNs with NFVs managing the RAN and the 5G network quality of service (QoS) will become important, especially in industrial use cases where accurate synchronization between machines is critical. Next-generation 5G networks will operate at more than 10X the speed of 4G networks. It is so fast that a 60-minute MP4 file of a TV show could be downloaded in fewer than 4 seconds. There are other ways to broadcast TV over 4G, so just an increase in network bandwidth won’t help the user or the service provider. But new low latency applications could be impactful as automation and IoT grow.
“This combination of technologies, robots and virtual reality will extend and augment human capabilities. Robots controlled with 5G could give a human greater strength for industrial applications or smaller size for telesurgery. A robot could be controlled over very long distances so that a robot could work in a hazardous place like the Fukushima nuclear reactor or a surgeon with specialized skills could perform an operation requiring very precise motion control.”
Increasing network speeds in the RAN is a partial solution. Engineering the networks as an SDN with NFVs will enable the networks to decentralize the traffic away from the core nodes where traffic is growing exponentially due to mobile data services, with the conditions threatening to become exacerbated with IoT and other emerging technologies.
Service providers can disproportionately scale their 5G networks this way by adding less equipment and capacity in the core than they do now by adding commodity hardware at the network edge for application processing, QoS and routing over the shortest path. This benefits service providers by reducing the capital investment in scaling their networks to carry increasing amounts of traffic.
The SDN can be thought of as a network hypervisor, interfacing to the network control plane and dividing up the RAN and network capacity into network slices and applying different levels of QoS. It is not too different from a VPN, but with the added capability of building applications that run dynamically and evolve over time without depending on the network equipment manufacturer.
NFVs are applications that run on the SDN hypervisor and process services created by the service provider, network software developers and users. Presumably, an ecosystem of application development will emerge, and there is a billing function like what Amazon has for billing CPU utilization, storage and services.
NFVs and the SDN will run on commodity hardware. NFVs provide computational capabilities for application processing, such as running application services that should run in the network, e.g., network synchronization and security features controlling which devices and systems can connect to others and in the case of IoT, to augment the devices with limited processing capability.
Fitzek’s demonstration includes an NFV that manages and coordinates low latency end to end. Each network element uses the NFV built for this demonstration. It controls its latency contribution to meet the QoS needed to run the application.
The point of Fitzek’s prototype is to demonstrate the 5G capability to control latency in the entire end-to-end network chain. It guarantees users an unprecedented level of QoS that is not possible with today’s technologies.
Given the possibility of creating new revenue streams from new use cases and reducing capital investment in core networks as data loads increase with mobile, wearables, IoT and other control applications, carriers are looking for and demonstrating what they hope will be killer apps while they deploy prototype 5G equipment in test beds.