At Orange Labs, Automated 5G Network Slicing Starts Now

[Oded]

Of all the new innovations 5G brings to mobile networks, none excites operators more than network slicing. Since the earliest cellular networks, there’s been basically just one kind of mobile service, limiting what enterprise customers could do with mobile connectivity—and the ways telcos could differentiate it. Network slicing turns that status quo on its head.

For the first time, mobile operators can slice up their networks into specialized sub-networks, each tuned to the unique needs of a specific service, application, or vertical. For example, one slice could be optimized for low-bandwidth, low-power-consumption Internet of Things (IoT) devices like sensors or power meters. Another could be dedicated to 4K video streaming. Still, another could be tuned for ultra-low latency communication with industrial equipment in a Smart Factory.

Start listing possibilities, and you can paint an amazing picture of what operators can do with 5G networks. But, you’re describing capabilities that go beyond what networks can support today. It’s natural to assume we’re talking about the distant future.

Well, get ready, because the future starts now.

HPE recently partnered with Orange Labs and Casa Systems to demonstrate what open, automated network slicing can look like in the real world. And here’s the best part: you don’t have to read through arcane technical reports to get the results. You can see them with your own eyes, in real-time.

Building tomorrow’s 5G services

If your goal is just to showcase a narrow 5G test case, it’s not hard to set up a proof-of-concept to do just that. HPE, and our partners at Orange Labs and Casa Systems, had more ambitious plans. We wanted to demonstrate slicing in a system that met the same kinds of requirements carriers will face in real-world networks. The test network had to be:

Open and multi-vendor interoperable. You can’t expect to support diverse vertical applications if everything you use has to come from one supplier. Operators need to be able to integrate 5G technology from multiple vendors, within a single framework.   

Automated end-to-end. Splitting a network into slices inherently adds complexity. Relying on manual configurations for all the technologies in the service path of each slice won’t work. You should be able to define what you need in a high-level orchestrator and have the network automatically instantiate the end-to-end slice, from the core through the RAN.

Elastically scalable. To monetize network slicing, you need to be able to guarantee those unique slice attributes under service-level agreements (SLAs). That means the network has to support closed-loop automation—adding resources to a slice or instantiating a new one in real-time to preserve SLAs when network conditions deteriorate.

It’s a tall order, but 5G operators can’t deliver network slices, much less monetize them, without those pieces in place. As we demonstrated in the video, these capabilities are not just science fiction. The tools for implementing open 5G network slicing exist right now.

Inside the Demo

In the video above, you can see the complete, cloud-native 5G test network we assembled at Orange Labs, using solutions from HPE and Casa Systems. We commissioned a fully-featured 5G Core with network functions from HPE and Casa. Then we used HPE’s Service Director to instantiate the network and initial slices, with predefined conditions (SLA violations) that would trigger an automatic response.

To simulate a mission-critical application, we chose a scenario that could easily be part of an operator’s portfolio: controlling a robot over a 5G network. This also allowed us to showcase network slicing in a way where it would be easy to see how it worked—or didn’t.

In the test, a small kit robot roams around a room, controlled by a remote operator over the 5G network. We then flooded the network with congestion. As you’d expect, the robot slowed down, clearly lagging in its response to commands. As soon as the robot’s responsiveness fell below an acceptable norm—violated the SLA we’d previously defined—the network sprang into action. HPE Service Director spun up another dedicated network slice, automatically configuring the proper quality-of-service and latency. The robot immediately regained its normal speed. And it all happened in seconds, without a human operator having to lift a finger.

Welcome to the 5G Future

Think of everything happening “under the hood” in this test: Location-based telemetry. Low-latency network slicing. Edge computing.

The test environment also addressed all the key requirements real-world 5G networks have to meet. This was a truly open, multi-vendor network, using cloud-native network functions from HPE and Casa Systems. It allowed Orange to configure a network slice end-to-end through HPE Service Director. And, it performed true closed-loop automation, dynamically responding to deteriorating network conditions to preserve an SLA.

This demo was, ultimately, a demo, not a live network serving real telco customers. But it should erase any doubts you might have about network slicing in open, multi-vendor environments: It’s real. It’s viable. And, it’s coming soon to a 5G network near you.

Interested to learn more about the architecture behind this?   read our new white paper: Reimagining Core Networks,  our visit our 5G knowledge sharing page: HPE.com/info/5G