Inferensys

Glossary

Guaranteed Bit Rate (GBR) Slice

A network slice type configured with dedicated network resources and a fixed bandwidth commitment, suitable for services requiring constant throughput like real-time video or industrial automation.
Stylish WeWork-like workspace with hot desks and document wall, professional searching through enterprise knowledge base on a mounted ultrawide display, warm industrial pendants overhead.
DEDICATED RESOURCE ALLOCATION

What is Guaranteed Bit Rate (GBR) Slice?

A network slice type engineered with a fixed, non-negotiable bandwidth commitment and dedicated radio and core resources to ensure constant throughput for latency-intolerant applications.

A Guaranteed Bit Rate (GBR) Slice is a 5G network slice instance configured with a hard resource reservation that permanently allocates a specific amount of radio, transport, and core network capacity to a tenant. Unlike best-effort slices, the Slice Admission Control function rejects new sessions if admitting them would violate the committed bit rate, ensuring deterministic throughput regardless of overall network load.

This slice type is essential for services like real-time industrial automation, professional video broadcasting, and V2X communications, where any fluctuation in bandwidth is unacceptable. The Slice-Aware Scheduler at the MAC layer prioritizes GBR traffic by assigning dedicated Physical Resource Blocks (PRBs) to these flows before serving non-GBR users, maintaining strict Slice SLA compliance for latency and jitter.

FOUNDATIONAL ATTRIBUTES

Core Characteristics of a GBR Slice

A Guaranteed Bit Rate (GBR) slice is defined by a set of non-negotiable performance characteristics that distinguish it from best-effort connectivity. These attributes ensure deterministic behavior for mission-critical applications.

01

Fixed Bandwidth Commitment

The defining feature of a GBR slice is a hard-coded minimum bit rate that is permanently reserved, regardless of overall network load. Unlike non-GBR slices that rely on statistical multiplexing, this slice type pre-allocates Physical Resource Blocks (PRBs) to satisfy the Guaranteed Flow Bit Rate (GFBR). This ensures that a remote-controlled robotic arm always has the exact uplink capacity required for real-time telemetry, even during peak hours in a congested cell.

Fixed
Resource Allocation Model
GFBR
Enforcement Parameter
02

Strict Packet Delay Budget

GBR slices enforce a deterministic latency profile defined by the Packet Delay Budget (PDB). This parameter specifies an upper bound for the time a packet can spend traversing the 5G system between the User Equipment (UE) and the N6 interface. For industrial automation, this is often configured to sub-5ms values, requiring slice-aware scheduling at the MAC layer to prioritize GBR traffic over non-GBR flows, preventing jitter that would destabilize a motion control loop.

< 5 ms
Target PDB for URLLC
03

Admission Control Enforcement

To protect existing sessions, a GBR slice implements strict admission control. When a new Protocol Data Unit (PDU) session requests resources, the Slice Admission Control function calculates if the remaining unreserved PRBs can satisfy the new request's GFBR without violating the guarantees of active sessions. If resources are insufficient, the request is rejected immediately, preserving the deterministic quality of service for all admitted users. This prevents the tragedy of the commons seen in non-GBR slices.

Reject
Action on Resource Shortfall
04

Maximum Bit Rate Capping

While a floor is guaranteed, a ceiling is also enforced via the Maximum Flow Bit Rate (MFBR). This parameter caps the throughput a session can achieve, preventing a single UE from monopolizing excess resources. This is critical for slice isolation—it ensures that a misbehaving application on one GBR slice cannot burst into the capacity reserved for another tenant's critical slice. Traffic exceeding the MFBR is shaped or policed at the User Plane Function (UPF).

MFBR
Throughput Ceiling Parameter
05

Energy Efficiency Tension

GBR slices present a fundamental challenge for energy-saving features like Cell DTX or Sleep Mode Coordination. Because resources are permanently reserved, the base station cannot easily deactivate carriers or mute resource blocks during low activity without risking a violation of the GFBR. Advanced energy-aware schedulers must therefore use predictive analytics from the NWDAF to identify genuine idle periods within the GBR allocation before triggering low-power states, balancing the SLA against the Slice Carbon Footprint.

High
Energy-Saving Complexity
06

QoS Flow Binding

Within a GBR slice, the actual guarantee is enforced at the QoS Flow level, not the PDU session level. A single PDU session can host multiple QoS Flows, each with its own GFBR and MFBR. The Service Data Adaptation Protocol (SDAP) layer maps these flows to Data Radio Bearers (DRBs). This granularity allows a single industrial sensor to simultaneously transmit a GBR flow for safety-critical shutdown signals and a separate non-GBR flow for firmware telemetry, all within the same slice instance.

QoS Flow
Granularity of Guarantee
GBR SLICE ESSENTIALS

Frequently Asked Questions

Clear, technical answers to the most common questions about Guaranteed Bit Rate network slicing, its architecture, and its role in energy-efficient 5G deployments.

A Guaranteed Bit Rate (GBR) slice is a network slice instance configured with dedicated, non-shareable radio and core network resources that provide a fixed, minimum throughput commitment to every admitted data flow. Unlike non-GBR slices that rely on best-effort delivery, a GBR slice enforces a hard Service Level Agreement (SLA) by reserving specific Physical Resource Blocks (PRBs) in the RAN and guaranteed bandwidth in the user plane function. The Slice Admission Control function rejects new Protocol Data Unit (PDU) sessions if accepting them would violate existing guarantees. This is achieved through slice-aware scheduling at the MAC layer, where the scheduler prioritizes GBR bearers over non-GBR traffic, ensuring deterministic latency and throughput suitable for real-time video, industrial automation, and Ultra-Reliable Low-Latency Communication (URLLC) applications.

Prasad Kumkar

About the author

Prasad Kumkar

CEO & MD, Inference Systems

Prasad Kumkar is the CEO & MD of Inference Systems and writes about AI systems architecture, LLM infrastructure, model serving, evaluation, and production deployment. Over 5+ years, he has worked across computer vision models, L5 autonomous vehicle systems, and LLM research, with a focus on taking complex AI ideas into real-world engineering systems.

His work and writing cover AI systems, large language models, AI agents, multimodal systems, autonomous systems, inference optimization, RAG, evaluation, and production AI engineering.