Network slicing is one of the essential elements of the full vision of 5G as an agile network which can be tuned to any user’s requirements in a dynamic fashion. Without slicing, it will be hard for 5G to meet the needs of many of the industries which hope to use it as part of their digital transformations. The demanding, and very diverse, connectivity requirements of sectors such as public safety, manufacturing, logistics or smart cities cannot be met optimally on a general purpose network, but the individual private network model, while important for some scenarios – especially in-building or localized build-outs – is hard to scale, and hard for MNOs to align with their core profit models.
Slicing, then, is essential, and while it has generally been seen as a rather distant goal, some vendors and operators are working to accelerate its adoption. This brings risks. In some cases, solutions are branded as slicing but support only a few static slices, which limits the monetization options. The fully dynamic configuration of slices on-demand will require the 5G core, and end-to-end connectivity which will include air interface slicing.
That means true slicing requires the deployment of a sophisticated iteration of the 5G core and some of the elements of the 3GPP Release 16 standards, just finalized, to ensure real time synchronization and other features. Operators in China, in particular, are leading the way to test large-scale dynamic slicing engines, but they acknowledge that such early adoption requires a large investment in customized solutions, which will not support the open frameworks that will, if they emerge, make slicing more affordable and scalable in future, and make it easier to create slices from resources owned by multiple operators.
Zhang Jing, a senior engineer at the China Information and Communications Technology Institute (CAICT), reflected on the risks at a recent network slicing conference in the country. He warned that early deployers would need to adopt immature solutions that would require a great deal of integration work, and that there was no interoperability between different vendors’ offerings as yet.
“There is still a certain gap between network slicing and commercial deployment,” Zhang said. There was “a long way to go”, and far greater cooperation was needed between all stakeholders.
Key problems are, Zhang said, that there are no commonly agreed approaches to key slicing processes, such as coordination between access, transport and core networks, or automated deployment of end-to-end slices.
“Each manufacturer implements its own private solution, and end-to-end network slicing capabilities have not been verified,” Zhang said.
Zhang is also a member of the China IMT-2020 5G Promotion Group, which is carrying out trials of end-to-end slicing in a program that will run from mid-2020 to the end of 2021.
And all three Chinese operators are working on trial or even commercial implementations of slicing, despite their reservations. Wang Changling, chief architect of China Unicom’s core network, echoed Zhang’s words, telling the conference that slicing “still has a way to go”. She outlined a project being run by Unicom, Huawei and appliance supplier Gree Electric to build a virtualized private network using edge cloud and 5G slicing; as well as a cloud gaming pilot, working with public utility State Grid and web giant Tencent
At China Telecom, Chen Yunqing, VP of the operator’s R&D function, said pilots of 5G slicing were running in six cities including Beijing and Shanghai, supporting applications such as live video broadcasting, remote monitoring, remote medical treatment and smart policing. The telco’s first commercial deployment will be a small 5G private network, based on a slice and edge computing, to support public safety applications for the police force in the city of Shenzhen.
China Mobile, despite being one of the pioneers of the slicing concept, is being even more cautious. It has various trials and pilots in process, but recently published a white paper that sets out the three stages it envisages in the journey to full 5G slicing. The third stage will take another 2-3 years to reach, the operator believes, but at that point, slicing will enhanced with AI-enabled service level agreements (SLAs) and dynamically self-optimizing networks which adapt on the fly to changing needs of the network and use case.