Network slicing will be an important enabler of the ability to execute on 5G business models. But how networks are sliced will have a big impact on how services are rolled out.
If the full potential of 5G is to be realized, networks will have to be sliced into discrete “layers” capable of supporting a range of requirements from how devices access the network to service-level agreement enforcement to security policies, etc. However, not all network slices are created equal. Even where an “end-to-end” network slice is required, the concept of horizontal vs. vertical network slicing will also be important.
Characteristics of a vertical network slice
As concepts related to 5G and the internet of things have become increasingly muddled, it can be tempting to think of network slicing use cases purely in terms of how they’ll be used to serve IoT requirements in a 5G world. However, while IoT will play a huge role in driving the need for network slices, it is far from the only driver. In fact, the need to serve multiple constituencies across a variety of use cases in specific vertical industries will be as important a network slicing driver as IoT.
In looking at these requirements, the case for vertical network slicing becomes more apparent. One simple example of a vertical network slicing scheme in action can be viewed through the lens of the automotive industry. On one hand, vehicle connectivity in a public safety situation is a major driver in the critical communications market. In order to ensure that public safety vehicles receive the proper prioritization in terms of ultra-high-speed data access for building schematics, local maps, communications, etc. a sophisticated set of networking requirements must be met and SLAs enforced. On the other hand, vehicle connectivity to consumer vehicles is subject to much lower standards with respect to real-time communications. In fact, in today’s connected vehicle market, the majority of infotainment content can be delivered via standard 4G/LTE connections, while other facets such as over-the-air software updates are generally tagged to take place over Wi-Fi connections at off-peak hours.
When examining both use cases, even at a high level, it is clear that each use case carries very different requirements. As such, in a vehicle-to-X-type network slicing scenario, within the same vertical network slice, two separate slices could be created to meet the different requirements of each constituency.
Characteristics of a horizontal network slice
Just as vertical slices can be used to serve different needs within a particular vertical, horizontal slices can be created support similar classes of devices, which could then be used to serve end users in multiple verticals. For example, a low-power network slice could be used to connect sensors from a variety of IoT-enabled devices. As a case in point, a slice could be created to serve the needs of consumer electronics in a home environment (i.e., smart appliances) as well as a smart-meter application for use in a smart utility situation. In this case, the network slice would be providing similar service characteristics, albeit to a potential diverse mix of customer segments.
A better approach? Horizontal, vertical or both?
Regardless of how network slices are characterized, one key takeaway from a horizontal vs. vertical network slicing discussion is the fact that network slicing schemes are going to have to meet a range of requirements, along a range of vectors, aimed at multiple industries. While this clearly points to the need for dynamic, on-demand network slicing capabilities, as a first step, network technology suppliers and network operator alike should consider first rolling-out pre-defined, generic network slices aimed at horizontal and vertical use cases.
While these templatized slices might not meet all the demands of every potential customer in given vertical or in support of a given use case, it could materially simplify the steps required in initial network slicing go-to-market initiatives. In turn, this more simplified approach could provide key learnings that will lead to incremental progress in the ability to create fully automated, on-demand network slices in the future.
