Automation is mandatory for 5G – the question is, how?

Autonomous networks aren’t just a goal for communications service providers (CSPs) – they’re mandatory to stay in business. At TM Forum Action Week, CSPs and their suppliers identified four 5G use cases that require network automation supported by artificial intelligence (AI) and machine learning. The next steps are looking for repeatable patterns in these use cases and figuring out how to deliver them with network and cloud resources that are self-configuring, self-optimizing, self-organizing, self-healing and self-evolving.

“Tomorrow’s network operations center will have only a man and a dog,” George Glass, VP of Architecture & APIs, TM Forum, joked with members attending a day-long workshop on autonomous networks. “The man will be there to feed the dog, and the dog will be there to make sure the man doesn’t touch any of the computing systems.”

Indeed, CSPs must work toward full automation of network and operational resources because it simply will not be possible to meet the latency and availability demands of 5G applications, such as autonomous vehicles, if provisioning, management and assurance processes remain largely manual as they are now.

“We need to completely overhaul the network to leverage the capabilities that are coming along from 5G, SDN [software-defined networking] and NFV [network functions virtualization],” Glass said. “If we don't, telcos will stay behind the curve, and these types of services will be delivered by somebody else.”

Increasing complexity

The trouble is, while the overall goal of automation is simplification, it actually increases complexity significantly. Whereas it used to take CSPs weeks or months to setup contracts with large enterprise customers for services such as virtual private networks, 5G applications necessitate that network capabilities scale on demand, in real time.

“With virtualized infrastructure, high-speed mobile access and hyperscale cloud data centers, I can set up a complex network configuration that connects multiple sites across the globe in less than 15 minutes…and it might be 15 seconds if everything was set up right. But to do this, we’ve got to take people entirely out of that process. There cannot be a human involved.”

TM Forum’s Autonomous Networks Project, began in May with publication of a white paper called Autonomous Networks: Empowering Digital Transformation for the Telecoms Industry. Members from BT, China Mobile, Ericsson, Huawei, Orange and Telstra collaborated on the paper and presented it at Digital Transformation World 2019 in May. Later, representatives from BearingPoint, Futurewei and TEOCO joined the team to progress the work. Here in Dallas, more than 30 companies attended the workshop to explore use cases and discuss next steps.

TM Forum members are working on automation as part of the Open Digital Architecture, which is part of the overarching Open Digital Framework. During the workshop in Dallas, Glass showed the graphic below to explain the goals of the project.

Autonomous vehicles illustrate the challenge

In a brainstorming session, the team identified about two dozen 5G use cases that require automation of resource management (for example, autonomous vehicles, connectivity as a service, gaming, smart factories and cities, remote surgery, and energy management). From the list they chose four to develop further through proofs of concept. They include:

• Connected vehicles – includes multiple scenarios ranging from autonomous driving to delivering on-line gaming to a passenger


• Smart cities – this could feature scenarios such as IoT sensor networks for real-time traffic management, energy management and assuring emergency services


• Smart manufacturing – scenarios could include rapid design changes, IoT sensors and monitoring worker safety


• Broadcast media – this could include instantiating pop-up networks for coverage of news or sporting events

Autonomous vehicles are often cited as the ‘holy grail’ of use cases for 5G and software defined networking (at Action Week on Tuesday, Shoma Chakravarty, VP of IT Enterprise Architecture, Verizon, highlighted this use case in discussing Verizon’s platform evolution). Because of their varied characteristics (low latency, high availability, machine to machine), connected vehicles do a great job of illustrating the requirement for automation and the complexity.

Eric Troup, Microsoft’s Chief Technology Officer, Worldwide Communications and Media Industries, explained the challenge: A car is traveling down a 10-mile stretch of highway at 60 miles per hour, and there is a millimeter wave cell tower every 400 meters. The car has an eSIM in it, and the eSIM session is moving from one cell tower to the next with a latency requirement of 30 microseconds. The car has 10 computers communicating via IoT applications running on the same edge infrastructure, communicating with the car manufacturer, an insurance company, a power company and governmental agencies like the Department of Transportation or the Environmental Protection Agency. A passenger in the backseat playing an online game. In addition, the car is self-driving car and uses vehicle-to-vehicle communications.

“In this case, we have a whole bunch of workloads, each with different priorities, that have to laterally shift across the network as the car moves,” Troup explains. “To run on a converged infrastructure, we need the capability to make sure one workload doesn't crash another. We can’t do this today. But in the future, due to the economics, it’s going to be a converged packet network and converged edge cloud infrastructure…and we will have to apply AI to achieve autonomous resource management.”

Troup stresses that his example is for a single vehicle moving along a 10-mile stretch of highway. “Now let’s image this is a city and you’re dealing with millions of cars all going in different directions. Can you imagine the complexity?” Troup asked. “So that’s why autonomous resource management becomes essential.”

TM Forum members will be developing Catalyst proofs of concept to explore how to implement automation and AI in the selected use cases. If you are a telco or enterprise interested in sponsoring a project or a vendor who would like to participate in addressing one or more of the use cases, please contact George Glass