Provisioning the Transport Architecture for 5G Network Slicing

The number of 5G deployments has ratcheted up. As of the end of August 2020, the GSA reported there were 118 5G networks among commercial operators and another 279 mobile operators carrying out trials. ABI Research estimates there were 234 million subscribers worldwide by the end of 2020. As these 5G commercial deployments grow in number, the focus is shifting to what services will be supported. “Network slicing” is a key concept enabling the potential of 5G. Importantly, network slicing can offer different service levels of network availability, throughput, latency, level of security, quality of service, and several other performance indicators. Mobile operators are exploring the potential of network slicing. Here is a snapshot:

• In January 2019, senior management of AT&T (U.S.) declared they would be using the virtualization of its radio and core network assets as a platform for network slicing.
• In March 2020, Singtel (Singapore) announced a network slicing trial with Nokia to automate the delivery of digital services, giving Singtel greater control of services across the network while addressing its enterprise customer needs.
• In mid-2020, TIM (Italy) began supporting static 5G slicing in which universities in Milan and Paris were connected. The operator intends to support full dynamic network slicing once greater network automation is place.

Network slicing, however, is not a 5G “free lunch.” Operators will need to move away from “best effort” to being able to offer prioritized packet data delivery and very low latency services that could be as low as 1 millisecond. This will be challenging even with 5G, but mobile operators will need to commit to service-level agreements (SLAs) which would have penalties for the mobile operators if not adhered to. Those SLAs may not be as strict as those for an end-to-end fixed telco network, but enterprise customers will expect them.

Mobile operators will need to ensure their telecom transport connections have sufficient capacity. Various backhaul technologies such as fiber-optic cable, E-band millimeter-wave backhaul links, and LoS MIMO and XPIC have boosted the throughput of individual links. However, the amount of data that can be supported per link has started to bump up against Shannon’s Law. Bonding interfaces is a practical solution to ensure there are no bottlenecks between the edge of the network and the core network. The ITU-approved Flexible OTN standard, “FlexO,” allows client OTN handoffs above 100 Gbps by operating a modular “OTUCn” approach, where “n” is the number of interfaces. FlexO also supports standard 100 GbE optical modules. Crucially, the ITU FlexO standard will not only support a very robust 100 GE throughput, it will also support existing OTN standards.

The other standard that is helping operators manage traffic on their transport backhaul connections is “FlexE.” FlexE improves end-to-end manageability, router-transport connectivity, and utilization. The Optical Internetworking Forum (OIF) promulgated the Flexible Ethernet (FlexE) standard to support bundling and multiplexing of interfaces between routers and transmission systems. FlexE, therefore, complements FlexO. FlexE offers a way to transport a range of Ethernet-enabled data rates irrespective of whether they correspond to the existing “physical Ethernet data rates” (i.e., some data packets may travel at a lower rate than the physical Ethernet medium can support). Furthermore, FlexE would allow Ethernet data rates to grow within an overall Time Division Multiplexing framework. This would allow network slicing where different bundles of services with different SLAs can “cohabit” on the same transport layer.

The tripartite features of ultra-fast broadband, ultra-reliability, and ultra-low latency will provide an opportunity for mobile telcos to evolve and take on web-scale type services and applications. Telcos will be providing ultra-fast and ultra-low latency links between their customers (be it prosumer or enterprise) and data center services (e.g., remote access storage and cloud applications) or even connecting their customers to edge cloud and computing functions that are directly supported by the telcos themselves. These edge cloud data centers may be deployed at certain localized macro cell sites spread across the country or more regionally at central offices (COs), or more to the center of the network at metro-centric data centers. How this edge cloud architecture will evolve is still up for debate, but it will be network slicing that provides the automation, management, and application layer attributes, while FlexO- and FlexE-capable transport-routers and switches from Arista, Ciena, Cisco, Ericsson/Juniper and Huawei ensure rapid routing over the transport layer.