Building 5G networks from the core up

-August 07, 2017

The world’s appetite for connectivity, bandwidth, and advanced high-speed, low latency networks is growing fast. The networking industry is in a race to stay ahead of the demand curve being set by consumers, enterprises, and government regulators.

5G will be a key technology service providers will use to meet customers' expectations and government policy goals, as well as to diversify beyond traditional connectivity into lucrative new business opportunities based on network performance characteristics not provided by 3G and 4G. Delivering that performance will require a different network approach not just for the access network but also for the next-generation core. The core will need to be designed and built using all the capabilities of the cloud, and adopt a more distributed architecture than today’s networks.

Figure 1 New user demands – with extremely diverse requirements

5G brings significant network improvements in capacity, connectivity, latency, and reliability. It will also need to support very diverse use cases with different requirements for latency, throughput, and availability. This will lead to a whole range of new services that can be broadly classified into three categories. The first is extreme mobile broadband. This takes what we are already familiar with as a mobile data service, but significantly boosts performance. It will handle new enterprise services and applications along with the exploding consumption of multimedia, such as augmented and virtual reality, and video in all its various forms and formats.

The second general trend is massive machine communication (MMC), also referred to as the Internet of Things (IoT). With the tremendous connectivity and scalability afforded by 5G, it will give rise to smart homes, cities, and the 4th industrial revolution, which will be characterized by smart factories. The third trend, which is part of MMC but more demanding, is the rise of critical communications requiring very high reliability and very low latency, for example, in the area of public safety.

Figure 2 Fourth industrial revolution powered by 5G

This diverse mix of services will create a range of demands and a level of unpredictability for which today’s networks have not been built. In order to address some of these needs, the 4G/LTE standard has already evolved through versions 4.5G, 4.5G Pro, and 4.9G (coming soon) along with unlicensed and shared spectrum technologies. To deliver the network requirements demanded by this broad range of services will require the continued evolution of 4G, unlicensed and shared spectrum technologies, and the introduction of 5G radio access, all anchored on a cloud-native, next-generation core.

To understand why this is so, it is helpful to look at the unpredictability that characterizes many MMC applications. Nokia Bell Labs estimates that the number of IoT-connected sensors and devices will be over 100 billion by the year 2025. This will be an order of magnitude higher than the number of end user devices, such as wearables, smartphones, tablets, and PCs. Many of these IoT devices, such as sensors, will be mostly passive and predictable, except for infrequent occasions when cascading events may lead to high and unpredictable levels of activity.

Imagine, for instance, an accident on a busy road filled with autonomous vehicles and the effect it might have on the complex meshes of sensors and devices both on the roadway and on the vehicles involved. Given the low latency requirements of autonomous vehicles, there would be an immediate need for massive scaling in data processing and responsiveness on a local level to support real-time and ultra-reliable vehicle-to-vehicle and vehicle-to-network communications.

Figure 3 Increasing safety with autonomous vehicles

This example illustrates a number of key requirements of critical MMC applications: ultra low-latency, reliability, and the need for network resources to be precisely placed where they are needed. Critical machine communications require very low levels of delay in network responsiveness or latency, in the order of 5–10ms. Thus, network resources for the processing of inputs from these devices (e.g. autonomous vehicles) and the triggering of critical automatic responses need to be at the edges of the network.

Reliability cannot be cost-effectively achieved by over-provisioning network resources to cover the worst-case scenario. This is where cloud-based designed core networks show their strength, as shared virtual network functions (VNF) can be re-allocated and spun up on-demand. The 5G network functions and architectures are being defined in such a way that the next-generation core can be modularized by function, enabling the virtual core resources for that service to be distributed and deployed as close to the end-user as possible.

As mentioned above, a key concept to the proposed 5G architecture is the full separation of control and user plane functions. Coupled with stateless VNFs that separate processing and storage, this enables underused processing resources, say on the user plane, to be used for control plane congestion typical of high levels of sensor activity, and vice versa. For example, the state of a failed control plane function can be instantly retrieved from a common data layer (database) and pushed to another active control plane function. This leads to more efficient use of network resources since the network functions are no longer duplicated just for service reliability. There is also support for the wider distribution of core resources with concurrent access to both local and centralized services so that user plane functions can be deployed closer to the user to meet lower-latency requirements.

As 5G services emerge, a cloud-native core has to be programmable, having the flexibility to be configured either in a stand-alone or non-stand-alone deployment option and either in a reference-point or services-based architecture to suit the needs of the service provider. It will be a truly converged, multi-service and multi-access core that supports all generations of mobile, unlicensed, and shared spectrum services, such as Wi-Fi and MulteFire, as well as fixed access technologies, whether copper- or optical-based.

Figure 4 Flexible network slicing

As new services get layered onto the network, a cloud-native core will have the ability to create an instance—or slice—of an entire network virtually, fully customized with dedicated network resources allocated by use case, subscriber-type, or application from a common infrastructure. This provides much needed security mechanisms to segregate enterprises or differing MMC-application-dedicated services from each other. It also ensures that service-level agreements (SLA) and quality of service (QoS) for that service are maintained.

The importance of this has been seen recently with several IoT security breaches. Market regulation of device security will be difficult to enforce, and as we’ve seen, hundreds of thousands of insecure IoT devices from a single negligent vendor can be used to wreak havoc. Network slicing enables secure management of IoT applications, isolating them from general, insecure activity. For instance, in the case of autonomous vehicles, network resources for the intelligent transportation system (ITS) communication functions between sensors and vehicles can be assigned their own, protected resources, although they are still taking advantage of the same common pool of cloud resources.

Network slicing will also be especially interesting to enterprises that are heavy users of IoT/MMC and critical MMC in their operations. They can create slices that isolate MMC traffic from general employee communications and data storage. They can also create slices that cross the LAN/WAN demarcation, creating secure LAN-like communications for employees operating anywhere. For enterprise operations that need to set up temporary operations, such as resource extraction, core services can be quickly setup and extended to support them.

This shift to a cloud-native core fortunately matches a concurrent transformation within most network operations toward the cloud with NFV/SDN, whether in service provider or enterprise organizations. They are already embracing the advantages that will come with this shift, such as cloud-agile operations and automated life cycle management of network functions. There will be cultural shifts needed as IT skills become more important and siloed operational teams learn to work more closely on the converged network core. Implementing the next-generation core is part of a larger transformation of telecom and network operations generally, as we prepare for the next wave of real-time, haptic, and critical-machine types of applications.

David E. Nowoswiat is a Sr. Product and Solutions Marketing Manager with over 25 years of telecom industry experience in both wireless and wireline technologies. He is currently in Nokia’s IP and Optical Business Group supporting cloud and 5G next generation packet core Marketing.

Also see:

Loading comments...

Write a Comment

To comment please Log In