5G Network Slicing

5G Network Slicing is a groundbreaking evolution in network design that enables service providers to deliver customized, virtualized, and isolated networks over a shared physical infrastructure. Unlike LTE, where all users share the same network characteristics, 5G introduces the ability to create independent logical networks (slices) – each tailored for specific use cases, performance levels, and service guarantees.

Each slice behaves as a dedicated end-to-end network, optimized for unique service requirements such as ultra-low latency for autonomous vehicles, high throughput for streaming, or massive connectivity for IoT deployments.

The Need for Network Slicing

The diversity of 5G applications cannot be fulfilled by a single “one-size-fits-all” network. From mission-critical healthcare systems to smart factories and connected cars, every use case demands a unique set of performance metrics.

Why Slicing Matters:

• Differentiated Services: Custom SLAs for each application type.
• Efficient Resource Utilization: Shared physical resources managed dynamically.
• Operational Isolation: One slice’s fault does not impact others.
• Monetization Opportunities: Operators can sell dedicated slices to enterprises.

Core Concepts and Terminology

Network slicing introduces several key terms that define how logical networks are created, managed, and identified across the 5G ecosystem. These elements form the foundation for orchestration, SLA enforcement, and interoperability between RAN, Core, and Transport domains.

Network Slice Instance (NSI)

An NSI represents a complete, end-to-end logical network running across the RAN, transport, and 5G Core. Each NSI delivers a specific service or tenant requirement—like a factory automation network, a video-streaming slice, or an IoT monitoring service. Every NSI operates as an isolated environment with its own resource allocation, QoS parameters, and management policies.

Example: A telecom operator might create three NSIs—one for public eMBB services, one for enterprise URLLC applications, and another for IoT sensor connectivity.

Network Slice Subnet Instance (NSSI)

Each NSI is composed of multiple NSSIs, which correspond to different network domains:

• RAN NSSI – manages radio resources, schedulers, and PRB allocations.
• Transport NSSI – defines routing paths, QoS queues, and SRv6/TSN configurations.
• Core NSSI – includes AMF, SMF, UPF, and PCF instances that control session management and policy enforcement.

This modular design allows independent scaling and management of each domain while maintaining end-to-end service integrity.

S-NSSAI (Single Network Slice Selection Assistance Information)

The S-NSSAI uniquely identifies a network slice for a user or service session. It is composed of two elements:

• SST (Slice/Service Type): Defines the nature of the service category.
• SD (Slice Differentiator): A 24-bit field used to distinguish between multiple slices within the same service type.

S-NSSAI = { SST , SD }

SST (Slice/Service Type)

The SST indicates the category of service provided by the slice:

SST Value	Slice Type	Description
1	eMBB	Enhanced Mobile Broadband – high throughput for data-intensive apps like AR/VR and video streaming.
2	URLLC	Ultra-Reliable Low Latency Communication – sub-10 ms latency for mission-critical services like autonomous vehicles or industrial robots.
3	mMTC	Massive Machine Type Communication – supports millions of low-power IoT devices over large areas.

Some operators extend SST values for private or custom slices (e.g., SST=128 for enterprise private networks).

SD (Slice Differentiator)

The SD provides additional granularity within the same SST. It ensures that multiple slices of the same service type (for example, two different eMBB slices for distinct enterprises) can coexist without conflict. The SD value, encoded in 24 bits, acts as a unique slice tag for operators.

Example:

• Slice A: SST = 1 (eMBB), SD = 0x101 → Consumer mobile broadband.
• Slice B: SST = 1 (eMBB), SD = 0x102 → Enterprise broadband with premium SLA.

Together, SST and SD empower 5G networks to dynamically recognize, select, and manage services per user, location, and application type.

End-to-End 5G Network Slicing Architecture

In 5G, network slicing enables the creation of multiple logical networks over a shared infrastructure, each optimized for specific service requirements like eMBB, URLLC, or mMTC. The figure shows how various 5G Core and RAN functions interact during PDU session establishment to ensure that each user or service connects to its assigned network slice with guaranteed QoS and policy control.

Key Flow Steps:

• UE initiates registration and slice request (S-NSSAI).
• gNB forwards registration to AMF via N2 interface.
• AMF authenticates the UE using UDM and queries NSSF for slice selection.
• NSSF returns the Allowed Slice (S-NSSAI) for UE connection.
• AMF triggers SMF to create a PDU session for that slice.
• SMF obtains QoS and policy from PCF and configures the UPF.
• UPF establishes user-plane paths and forwards data to DN via N6 interface.
• All interfaces (N2–N6) ensure end-to-end slice isolation, QoS, and traffic control.

Lifecycle Management

The lifecycle of a 5G network slice is defined by 3GPP specifications TS 28.530–28.541, which outline the complete process from design to termination. It ensures that each slice is created, managed, and retired systematically while maintaining service quality, resource efficiency, and SLA compliance across RAN, Core, and Transport domains.

The six key phases are:

• Design: Define the service-level agreement (SLA), coverage area, latency targets, and resource allocation policies. This phase translates business and performance requirements into technical slice blueprints.
• Instantiation: Deploy and configure the necessary sub-slices across different domains -RAN, Core, and Transport. Network functions and virtual resources are allocated and linked together to form the end-to-end slice.
• Activation: Activate the slice by assigning resources, establishing control and user plane connections, and enabling the defined services for users or enterprises.
• Operation: Continuously monitor network KPIs, enforce QoS dynamically, and ensure SLA adherence using analytics and automation (e.g., via NWDAF and policy control).
• Modification: Scale or adjust the slice configuration based on traffic load, user demand, or SLA changes, adding capacity, reallocating resources, or updating policies as needed.
• Termination: Gracefully release all allocated resources when the slice is no longer required, ensuring clean de-registration and minimal service impact on shared infrastructure.

Core Network Slicing

The 5G Core introduces control and user plane separation (CUPS) to manage multiple slices efficiently.

Control Plane:

• NSSF: Determines slice availability per region.
• AMF: Selects appropriate slice during registration.
• SMF: Establishes and maintains per-slice PDU sessions.

User Plane:

• UPF: Slice-specific data forwarding and QoS enforcement.
• Edge UPF: Deployed closer to the user for low-latency services.

QoS Flow Management: Each slice supports multiple QoS flows identified by QFI, mapped to DRBs through SDAP at RAN.

Transport and Orchestration Layer

The transport network ensures slice isolation across IP, MPLS, or SRv6 layers. Orchestration is handled via NFV-MANO and ONAP frameworks.

Techniques:

• Segment Routing (SRv6) for slice-aware paths.
• Time-Sensitive Networking (TSN) for deterministic services.
• Software-defined control for bandwidth slicing.

Security and Isolation

Each slice enforces strict security separation:

• Dedicated or logically isolated AMF/SMF/UPF instances.
• Slice-specific encryption and authentication.
• Lawful intercept compliance per slice.
• Secure APIs exposed via NEF (Network Exposure Function).

Key KPIs and Performance Targets

Parameter	eMBB	URLLC	mMTC
Peak Data Rate	20 Gbps	1 Gbps	100 kbps
Latency	20 ms	1 ms	100 ms
Reliability	99.90%	100.00%	99%
Device Density	10⁴/km²	10³/km²	10⁶/km²

Testing & Validation

Validation of slicing requires end-to-end verification using KPIs and real-time logs:

• Functional Tests: Slice selection, registration, mobility.
• Performance Tests: Throughput, latency, jitter under load.
• Interoperability Tests: Multi-vendor gNB ↔ Core Slice validation.
• Resilience Tests: Slice recovery under failure or overload.

Example Deployment Scenario – Smart Campus

Enable an enterprise campus with three dedicated slices.

Slice	Use Case	SLA	Deployment
URLLC	Robotic Automation	<10 ms	Edge UPF + TSN transport
eMBB	AR/VR Learning	>1 Gbps	Central UPF
mMTC	IoT Sensors	10 kbps/device	Cloud UPF + IoT core

The Road Ahead: 5G-Advanced and 6G

• AI-driven slicing: Predictive scaling using NWDAF and ML models.
• Green slicing: Energy-efficient resource allocation.
• NTN integration: Satellite-based slices for global coverage.
• Dynamic slice orchestration: On-demand slice creation via APIs.

5G Network Slicing is the technological backbone of service differentiation in next-generation networks. It enables operators to evolve from connectivity providers to digital service enablers, offering customizable, on-demand virtual networks with guaranteed performance. As 5G transitions to 5G-Advanced and 6G, slicing will become the key enabler for industries such as autonomous vehicles, smart manufacturing, and AI-driven IoT ecosystems.

References

• 3GPP TS 23.501 / 23.502 / 23.503 – 5G System Architecture
• 3GPP TS 28.530 / 28.531 / 28.541 – Slice Management Models
• ETSI NFV-MANO Specifications
• IETF DetNet and SRv6 Drafts