When meeting customers, I often hear them comment that seeing network infrastructure as a dumb pipe (assuming intelligence will be provided by applications) is no longer valid. CSPs are looking for more and more intelligence in the network so that it can be fine-tuned to end-customer use-cases needs. In other words, it is not only applications and end-customer devices that will enable use cases; dynamically configuring connectivity as per use-case needs will also play a major role.

The kind of use cases these customers have in mind are enterprise applications such as collaborative robots in Industry 4.0, V2X communications for vehicles, including autonomous cars, truck platooning in transportation, connected ambulances in healthcare, remote learning in education, and fixed wireless access for enterprise connectivity. 5G network slicing is the kind of intelligent connectivity feature that these customers have in mind, because it enables connectivity with deterministic characteristics for bandwidth, latency and security that can be guaranteed by an SLA (service-level agreement) and can ensure the suitability of the end-to-end connectivity service for these kinds of use cases.

Customer remains the king

5G makes it possible to address enterprise customers’ needs as never before.  

Their expectations have evolved and are quite different compared to consumers. The main use case for consumers is still enhanced mobile broadband (eMBB). Enterprises use eMBB, but are also looking for ultra-reliable low-latency (URLLC) and massive machine-type communications (mMTC) to meet the needs of their advanced use cases. Guaranteed SLA-based connectivity will be a must.

To serve the needs of specific connectivity characteristics, there are two key approaches:

1. The enterprise builds its own 5G private wireless network for addressing its specific needs,
2. Network slicing provided by a slice provider (this could be a CSP or other third-party service providers) to slice tenants, such as enterprises, based on an agreed SLA.

Forecasts for market adoption from ABI research, report that the total addressable market (TAM) for 5G in enterprise vertical domain will be $82Bn by 2030. Out of this, 5G private wireless is estimated to be 78 percent, whereas, for network slicing services, it is estimated at 22 percent. This isn’t uniform across all enterprise segments, small-to-medium businesses (SMBs) are more likely to choose the second option than large enterprises. 5G private wireless is better aligned to the security and customization needs of a larger enterprise, while network slicing services align to an OPEX-driven business model, with lower TCO, which is more attractive to SMBs.  

According to another ABI research, among different industry verticals, most of 5G customers are coming from manufacturing, logistics and automotive sectors.

Slicing challenges

• The slicing capabilities required by the CSP will depend on the customer’s use-case requirements; thus, selecting which use case to prioritize is crucial. Analyzing the business cases of different use cases in terms of when the CSP will see a return on the investment is complex. Especially because they must align their analysis to the market uptake of these services, which can be tricky, given these are early days for the adoption of many of these use cases.  In addition, CSPs are “doing their own math” on ROI, investments and looking at markets, selecting use cases as per business case.
   
• Slicing lifecycle management (LCM) requires sufficient NFVI infrastructure and capacity in the underlying network, including sufficient radio coverage, to support it. CSPs must have the flexibility to address capacity and coverage requirements as a prerequisite for automated slice LCM operations.
   
• Implementing many of these use cases will require an ecosystem of 5G devices be available, which isn’t always the case today. Along with devices, applications also need to be able to invoke specific capabilities from the network such as reliability, redundancy and ultra-low latency to be delivered by the underlying  end-to-end slice-based connectivity.
   
• Addressing the above is doubly challenging because 5G standards from 3GPP are still evolving.  For example, connected car applications are evolving at this point, although they will be deployed on an underlying ultra-low latency (URLLC) 5G network or URLLC slice on a 5G network. As 3GPP standards mature, more and more capabilities are being added and their implementation will result in more and more enterprise use cases becoming commercially available and viable.

A step-by-step approach

Now, we know that slicing will address enterprise use-cases requirements – the next question is how? This means answering the following questions (not an exhaustive list):

• Which use-cases to deploy in the network?
• How many slices shall we create?
• Where to enable these slices?
• What are the characteristics of the slice?
• What should the capacity and coverage be for each slice?
• How do we align network resources to business needs?

For CSPs contemplating network slicing, what is needed is an analysis of the requirements based on which services will be delivered by the use case. These requirements can then serve as inputs to slicing design, which addresses slice orchestration, assurance and automating LCM. Design also identifies the KPI criteria.

Given the complexity of the various layers of the network (e.g., core, RAN, transport, etc.) addressing the needs of slicing, and with each subnet having their own plan to introduce slicing features, end-to-end slicing enablement is more complex, making automation and agility in slicing operations more challenging to achieve.

In order to address these challenges, an incremental slicing approach works best, where the initial phase develops a generic set of slicing capabilities or slicing platform, which ensures end-to-end procedures for slicing orchestration, assurance and operations in an automated fashion. Subsequent phases can use this generic capability or platform to create specific slice instances based on use-case requirements of slice customers.

A slice instance is nothing but instantiation of a specific slice. For example, we can have multiple eMBB slice instances like eMBB1 with eMBB slice characteristics as 500 Mbps throughput, or eMBB2 with 300 Mbps throughput etc.

Following are the steps typically followed in ensuring generic slicing capability:

1. Network slicing design
2. Slice orchestration
3. Slice assurance
4. Slice LCM operations continuously addressing slice additions, deletions and modifications based on operational and business feedback.

Conclusion

For enterprises looking at SLA-based connectivity for their use cases, 5G private wireless and network slicing are the two key solutions. Still, there is a need to have a robust, end-to-end ecosystem including devices, applications, slice-based connectivity and OSS/BSS (Operations Support System and Business Support System) to ensure the successful deployment of enterprise use cases.

Slicing technology is complex and touches all domains (RAN, transport, core, etc.). To address the complexity, an incremental approach works best, where the preferred approach is to build a generic capability first as a platform and then use this platform for specific slice requirements for multiple customers which are nothing but network slice instances in the deployment.

Nokia offers products in the slice management domain as part of its end-to-end 5G product and solution portfolio. We also offer comprehensive professional services addressing slice consultancy, slice design, build and management. See our website for more information and our technical video introducing how Nokia automates deployment of 5G network slices, focusing on the 5G Core.