Network Slicing

1. Different Service Requirements and a Uniform Network

From their earliest days, mobile networks developed as shared infrastructure serving one primary service profile for all users. 2G and 3G networks were built for voice calls. The 4G LTE network was engineered primarily to provide fast mobile broadband to smartphones. This "one-size-fits-all" approach was highly successful and powered mobile app growth, but it limited the network's ability to match different service requirements.

The requirements of different applications vary. Streaming a 4K movie demands high bandwidth. Controlling a factory robot requires near-instantaneous, reliable commands. A smart water meter may send a tiny data packet once a day, while a suitable device design can target decade-long battery life. The single, homogenous 4G network architecture could not efficiently cater to these different, and often contradictory, needs. Providing an ultra-low latency connection for a robot would mean designing the entire network for that purpose, making it inefficient for high-bandwidth video streaming.

This is the problem that Network Slicing was designed to solve. It is a core 5G capability that lets a mobile operator support different service profiles on shared infrastructure. The operator can partition physical resources into multiple customized virtual networks. Each is called a "network slice," and its resources, policies, and characteristics can match a particular service, application, or customer. The degree of isolation depends on the deployment and on which functions are shared.

2. Network Slice Definition

A network slice covers more than a simple Quality of Service (QoS) profile. It is an end-to-end logical network created and managed for a service requirement. It can use functions shared with other slices while retaining its own policies, identity, and resource parameters.

End-to-End Scope

The end-to-end scope means that a network slice extends beyond a core-network feature. It is a continuous, logical partition that spans the communication chain:

• The User Equipment (UE): A smartphone or IoT device may include a preferred Requested NSSAI in registration if it supports slice selection. The network checks subscription, policy, and availability, then assigns an Allowed NSSAI.
• The Radio Access Network (RAN):5G base stations (gNBs) coordinate radio resources with slice requirements, for example through suitable bearers, priorities, or numerology. The exact configuration depends on the operator's implementation and policy.
• The Transport Network: Depending on the deployment, backhaul and midhaul connections linking radio sites to the core network may also need to differentiate and prioritize traffic belonging to different slices.
• The 5G Core Network (5GC): This is where the main logic of slicing is implemented. Each slice can have its own dedicated set of virtualized Network Functions, or it can share functions with other slices while having its own policies and user plane resources.

Isolation and Service Parameters

Each network slice is designed as a logically distinct network, and isolation limits the impact of traffic and incidents between services. A traffic surge or security issue in one slice, such as a distributed denial-of-service attack targeting IoT devices, can be contained by resource and security policies, while the outcome depends on shared functions and implementation. Defined isolation policies support measurable Service Level Agreements (SLAs) for each slice. The isolation level is a design and management property of the deployment.

Customization lets an operator define for each slice:

• Topology: Which network functions are included and how they are connected.
• Performance Characteristics: Target or contracted bitrates, maximum latency, jitter, reliability levels.
• Security Rules: Custom encryption and authentication policies.
• Access Control: Defining which users and devices are allowed to connect to the slice.

3. Service Profiles Supported by Network Slicing

Network Slicing helps one 5G infrastructure support the three scenarios described by the ITU: eMBB, URLLC, and mMTC. The profiles below are a teaching model, and operators choose parameters for the service. In 3GPP, standardized SST values include eMBB, URLLC, and MIoT, while mMTC is the IMT-2020 scenario name.

4. Architecture for Network Slicing

Network slicing relies on 5G architectural principles, particularly the 5G Core network and its service-based functions.

• Service-Based Architecture (SBA):The SBA's modular design fits slicing well. A slice uses Network Functions that may be dedicated or shared, with policies and resources assigned to the service. For example, a low-latency URLLC slice might use a different instance of the Session Management Function (SMF) and User Plane Function (UPF) than a standard mobile broadband slice. SBA service discovery supports creating and managing these configurations.
• Network Function Virtualization (NFV): Because NFs are implemented as software, NFV can let an operator instantiate SMF or UPF instances for a new slice when the service policy requires them, and release resources when demand ends. Dynamic allocation of compute and storage supports slicing elasticity.
• Software Defined Networking (SDN): SDN supports orchestration by centralizing part of the network control logic and resource view. Controllers and an orchestrator coordinate the radio, transport, and user-plane infrastructure to enforce slice traffic policies. The programmable scope depends on the domain and interfaces in use.

Slice Identification and Selection

A user device may include a Requested NSSAI with an identifier called the S-NSSAI (Single Network Slice Selection Assistance Information)in its signaling. S-NSSAI contains an SST and an optional SD. During registration, the network checks subscription, PLMN policy, and availability; the AMF coordinates the selection of an Allowed NSSAI and the resources for the user's session.

5. Operator Service Models

Network Slicing can broaden the business model for mobile operators. An operator can combine connectivity with a platform for specialized services whose parameters are described by policies and SLAs.

• Enterprise Services: Operators can offer standard mobile broadband slices to consumers, while tailored slices can serve enterprise and industrial customers.
• Network as a Service (NaaS): An operator can sell a data plan together with a virtual network slice and agreed performance parameters. For example:
  • A logistics company could purchase a nationwide mMTC slice for its asset tracking sensors.
  • A television broadcaster could lease a high-throughput eMBB slice with a guaranteed uplink bitrate to cover a major sporting event.
  • A city government could operate a URLLC slice for its public safety first responders, reducing the risk of communication disruption during a crisis.
  • A factory could deploy a private URLLC slice on its campus for its automated robotics, keeping all latency-critical data on-site.
• Additional Revenue Opportunities and Innovation:This ability to offer differentiated, SLA-backed services can create additional revenue opportunities for operators. It can support entry into vertical markets such as automotive, healthcare, and manufacturing, where connectivity is one part of a customer's digital transformation.

Network Slicing is an architectural capability that extends 5G beyond a uniform 4G-style service profile. It can configure shared infrastructure with flexible, dynamic, and programmable profiles for diverse connectivity requirements in successive deployments.