Big steps on the road to 5G

Not simply another increment in our mobile radio evolution, 5G represents a paradigm shift for the industry. Virtualization, software-defined networking, a converged packet core, new spectrum and hybrid access are some of the big steps the industry is taking on the road to 5G. One big step concerns backhaul, which is evolving from a static, linear connection to a programmable mesh interconnecting all mobile and cloud elements dynamically.

5G drivers
Three applications and services are the principle drivers for 5G adoption at this point in time. Immersive video tops the list, including augmented and virtual reality. Second is ultra-dense IoT in urban areas, which is expected to reach densities of up to one million IoT connections per square kilometer, many of them wirelessly connected. Wireless service delivery for what is being called Industry 4.0 is the third driver for 5G.

Industry 4.0 is the idea of smart manufacturing, a term which covers automated processes both inside and outside what we have traditionally called the factory. Of special concern for 5G is meeting the needs of latency-sensitive automated processes in widely distributed and often remote areas, such as occurs in resource extraction. The latency requirements for these automated processes are often less than half a millisecond, which requires certain functions of the packet core to be well within 100 km of the radio. This will require a new architecture for the packet core.

Of the three use cases, Industry 4.0 is perhaps the best one to illustrate dramatic changes with 5G. Never has the mobile network had such stringent demands placed upon it. These will be mission-critical applications where failure will not only be expensive and disruptive, it may also threaten the safety of workers. Industry 4.0 will not only demand a new packet core architecture that supports distributed core functions, but highly flexible access to adjust to rapidly shifting industrial requirements.

One of the key characteristics of mobile backhaul to date has, ironically, been its static, non-mobile nature. Cell towers don’t tend to move. Backhaul or fronthaul access to the tower, once in place, rarely has to be touched. In contrast, industrial applications relying on wireless connectivity to support automation, will be quite different. The entire supply chain of the Industry 4.0 will be highly dynamic and responsive to shifts in market demand.

This demand for almost instantaneous reconfiguration of manufacturing infrastructure must be matched by the network. This can also be seen in other IoT areas, such as autonomous vehicles. Imagine the mobile network responding to events such as freeway-blocking accidents with thousands of IoT-connected cars queueing within existing cells, or being re-routed over less-used rural roads.

Event-driven access demand is a good way to understand the new paradigm. From spontaneous street celebrations with the usual video sharing, to the rapid escalation of resources and information that accompanies increasingly severe climate events, the wireless network, including access, has to become elastic and dynamic. Over-provisioning access in anticipation of such events will not be a profitable strategy.

Centralized processing
Given the physical limitations of spectrum, 5G will also evolve to smaller cell architectures, both to accommodate denser broadband demand as well as the shorter wavelengths of new spectrum, such as mmWave. An issue with denser, smaller cells, however, is an increase in cell edge interference. Managing this interference is a big driver for centralized Radio Access Network (C-RAN) architectures where baseband processing is centralized. Pooling processing functions in hubs is an effective way to improve multi-cell interference management, ensuring the most spectrum is available for payable traffic.

Next-generation fronthaul will improve on C-RAN by redistributing the baseband unit (BBU) functions between the cell site and centralized hub locations. Current fronthaul implementations, which we see with C-RAN today, carry a great deal of pointless traffic related to sampling, which the BBU does with the antennas at the cell site. This is close to 90 percent of the traffic now being carried by fronthaul links back to centralized hubs.

In the next-gen C-RAN implementation, sampling would be handled at the cell site, as it is classically. Called “split processing”, the actual split of functions between the BBU and the centralized pool is currently being debated within the standards organizations, but the goal is to reduce the fronthaul bandwidth requirements by nearly an order of magnitude.

Centralized processing is a natural step toward cloud RAN and the use of cloud compute infrastructure to scale and reduce operational cost. Hosting centralized processing on cloud compute infrastructure means operators can respond faster to the scaling, performance and cost challenges. Imagine, for instance, operators hosting specialized, ultra-localized services for a long weekend event, such as a Formula 1 race.

Cloudification will help to support the expansion of IoT devices , whose data transmission may be small, but whose computational requirements are not, given the potential for millions and millions of devices .

To be clear about what we mean by dynamic, event-driven access, data centers and cloud compute infrastructures must already be fully interconnected by fiber and microwave links. Physical connectivity is never likely to be dynamic. What is dynamic is the packet network upon which a network service runs. This end-to-end network service needs to be instantiated appropriately to interconnect the mobile and cloud elements that provide the radio, RAN processing and packet core functions.

This is possible and is performed today to provision, assure and optimize network resources to deliver connectivity services with strict Service Level Agreements (SLAs). The automated workflow begins by having the orchestration system that establishes mobile applications also tell the Software Defined Network (SDN) controller to create the network service at the same time. As SDN was first pioneered in data centers, ’carrier SDN’ is used to describe the functionality needed to provision end-to-end network services over multi-layer and multi-technology wide area networks.

With network services being managed by the carrier SDN controller, the physical transport infrastructure becomes more of a programmable fabric which interconnects all mobile and cloud elements dynamically. This ‘smart fabric’ extends end-to-end and supports connectivity among virtualized processing and applications wherever they may be located. It uses the transport technology and network service type that makes the most economic sense.

The road to 5G
This dynamic, cloud architecture will provide 5G radio site access, as well as legacy radios, with full solution sets in all transport technologies to ensure the appropriate, most economical decision can be made. It will carry more traffic volumes and more traffic types, including low latency. It must also meet the perennial needs for synchronization, reliability and security. It will consist of a programmable, multi-layer and multi-technology fabric that supports SDN workflow automation and agile network services instantiation.

There are, as we’ve seen, significant application drivers for 5G. The industry is moving quickly and big steps are already being taken to get us there. The paradigm shift that 5G represents will require significant changes in the way transport networks are designed and operated, and these changes will affect all transport technologies. Whether it be backhaul, fronthaul or any other kind of hauling, we will need to take these steps quickly.

About the Author

Jim Guillet, Senior Director for Product Marketing at Nokia

Jim is part of the Nokia Marketing team based in Ottawa, Canada. He studies wide area networks and their applications for communication service providers, webscale and public sector organizations. An engineer by training and a story teller by trade, he loves stitching together insight and education to support meaningful conversations with customers. When he’s not working, you’ll find him trying to cook the perfect hamburger or discussing the local hockey team with his neighbors.