Exploring the Building Blocks of Network Slicing

It is not news to say that network slicing will be a key enabler for operators to extend their basic network connectivity to new revenue diversification models. Network slicing will allow operators to run multiple logical networks on a commonly shared infrastructure spanning multiple domains (radio, transport, core, and edge). This arrangement marks a departure from the unpredictable connectivity of legacy infrastructure by promoting flexibility and dedicated capabilities tuned to different enterprise use cases. Operators can cordon off a “slice” of the network by earmarking it according to specific criteria that would cater to service types across Enhanced Mobile Broadband (eMBB), Massive Machine Type Communications (mMTC), and Ultra Reliable Low Latency Communication (URLLC).

Managing the influx of unique Quality of Service (QoS) offerings in the 5G era, however, requires an upgrade in a network’s wireline infrastructure capabilities. “Soft” network slicing through segment routing has been key in supporting a diverse selection of service offerings, through its ability to simplify network protocols to deliver differentiated Service Level Agreements (SLAs) in an efficient manner. “Hard” network slicing, on the other hand, aims to establish highly dependable network infrastructure to constantly maintain separate network slices in networks that are serving multiple heterogenous QoS requirements for various applications.

In a soft network slicing scenario, the added QoS diversity requires assigning unique QoS Class Index (QCIs) values to each traffic class per device. The respective components of the network (Radio Access Network or RAN, backhaul, core) would then accommodate the stipulated QCI value of the traffic types by properly allocating adequate network resources to successfully achieve the latency/throughput requirements of said QCI value. The traffic types of the 3 main connectivity scenarios of 5G (eMBB, URLLC, and mMTC) have a standardized QCI value or Standardized Slice Type (SST) associated with them. These SSTs provide the baseline requirements for the respective network components to fulfill the unique scenarios.

Aside from ensuring differentiated treatment of QCI across different traffic types (eMBB, URLLC, and mMTC), the network would also need to have the ability to differentiate QoS at the customer level (referred to as hierarchical QoS). This is necessary for operators aiming to provide differentiated services. This feature is especially pertinent to the backhaul part of the network, which handles the aggregation of traffic types across multiple customers that have varying levels of latency, security and availability assurances in their SLAs.

The foundation of an effective soft network slicing execution is having more consistent tunneling protocols and handoff between all the components of a network to properly handle the different types of data traffic in a 5G network. Addressing these issues has led equipment vendors and operators to rely on segment routing for 5G soft network slicing. Segment routing can replace existing GTP-U tunneling protocols that legacy networks use to treat traffic from the base station to the core. The General Packet Radio Service Tunneling Protocol over UDP/IP (GTP-U) tunneling protocol is what ensures the different treatments of traffic types in a network. The drawback of this current protocol is that it is not well-equipped to overcome the fragmented connections across the RAN, backhaul, and core. A segment routing IPV6 (SRV6) enabled network, however, has the capability to overcome this fragmentation and make a consistent connection across different network domains. This consistent domain connection—along with segment routing’s traffic engineering, programmability, and service chaining—is a testament to both its superiority over GTP-U and its effectiveness for 5G networks.

Hard network slicing through FlexE is more straightforward, in the sense that it is more about instilling proper physical separation and treatment of different types of traffic, guaranteeing and prioritizing the QoS requirements of certain use cases. FlexE’s hard channelization is able to address the transition of an operator’s network from 4G to 5G by supporting various traffic types from different network generations (CPRI, eCPRI, Radio Over Ethernet, ORAN), allowing for both a forward-looking and backwards- compatible network. Additionally, FlexE is especially useful for SLA-based services that mandate true separation of traffic and assurance of low latency. It is also useful in edge sites that might require collocation of mobile network functions and enterprise network functions—a highly probable outcome, given the economies of scale and the business case for operators in designing their network offices to accommodate third-party providers. The hard separation of FlexE assures the security and QoS of each network slice in the presence of other, third-party traffic.

Another version of hard network slicing is oriented around logically grouped virtual components into what are effectively separate network instances (or hard network slices) that can then each be configured to meet the requirements of different use cases. This is done through the virtualization and disaggregation of network components, i.e., RAN (RU, DU, CU) or core network Control and User Plane Separation (CUPS). This allows operators to ensure QoS requirements through the flexibility of network architecture deployment that virtualization and disaggregation provides. CUPS allows for the geographical distribution of multiple UPFs to the network edge, ensuring the latency requirements of certain use cases; meanwhile, operators can also flexibly place the RUs, DUs, and CUs at strategic locations of their network to ensure robust fronthaul and midhaul latency rates.

Operators are increasingly seeking to create services that are more differentiated in a bid to generate new revenue sources and be more attractive to their enterprise clientele. From the operator’s perspective, network slicing affords the opportunity to be agile, thereby enabling the ability to capture new revenue streams from personalized SLAs. The increased service agility, improved economies of scale (through executing multiple slices on a common shared network infrastructure), and, most crucially, revenue generation, are highly dependent on a transport network that is capable of handling the scale and complexity of network slicing.

Network equipment vendor Ericsson has shown promise in its recent RAN slicing product. Ericsson’s RAN slicing feature can assign resources in the air interface depending on the QCI value of the traffic stream. Additionally, Ericsson’s Dynamic Radio Resource Partitioning (RRP) feature can provide separation between different traffic groups and secures the network performance of the respective traffic group to meet SLA requirements. Dynamic RRP, as its name suggests, also allows for dynamic resource sharing among multiple slices without statically reserving them. Any freed-up resources can be utilized by other slices that might need them  to prevent a deterioration of QoS from resource hogging an unused slice.

Intelligent solutions like these that can efficiently manage the increased quantity, scale, and service delivery of different slices (with reduced CAPEX and OPEX to boot) will arguably be the ultimate arbiter in how operators can monetize their network slicing offerings.