Inferensys

Glossary

New Radio Unlicensed (NR-U)

New Radio Unlicensed (NR-U) is the 3GPP standardized technology enabling 5G New Radio operation in unlicensed and shared spectrum bands, such as the 5 GHz and 6 GHz bands, while ensuring fair coexistence with Wi-Fi and other technologies.
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.
5G SPECTRUM SHARING PROTOCOL

What is New Radio Unlicensed (NR-U)?

New Radio Unlicensed (NR-U) is the 3GPP-defined technology standard enabling 5G New Radio operation in unlicensed and shared spectrum bands, such as the 5 GHz and 6 GHz bands, while mandating fair coexistence with incumbent technologies like Wi-Fi.

New Radio Unlicensed (NR-U), standardized in 3GPP Release 16, extends 5G NR into the globally available unlicensed spectrum. Unlike LTE-LAA, which required a licensed anchor carrier, NR-U supports standalone operation in unlicensed bands, making it a viable technology for private 5G networks and neutral host deployments. The core technical challenge is ensuring fair coexistence with Wi-Fi and other technologies through a channel access mechanism.

NR-U employs Listen-Before-Talk (LBT) as its primary coexistence protocol, requiring a transmitter to sense the channel and verify it is clear before transmitting. To further enhance fairness, NR-U introduces a contention window adaptation mechanism and supports wider bandwidths (up to 100 MHz) than Wi-Fi. The standard also defines discovery reference signals and flexible frame structures optimized for the bursty, asynchronous nature of unlicensed spectrum access.

5G IN UNLICENSED SPECTRUM

Key Features of NR-U

NR-U extends 5G New Radio into the 5 GHz and 6 GHz bands, enabling operators and enterprises to augment licensed spectrum with globally available unlicensed resources while ensuring fair coexistence with incumbent technologies like Wi-Fi.

01

Listen-Before-Talk (LBT)

The foundational coexistence mechanism mandated by global regulations. Before transmitting, an NR-U gNB or UE performs a Clear Channel Assessment (CCA) to sense the channel's energy level. If the channel is busy, the device defers transmission using a random backoff counter. This ensures NR-U does not monopolize the medium and shares fairly with Wi-Fi and other technologies. 3GPP specifies multiple LBT categories, from no LBT (Cat 1) for short control transmissions to full random backoff with variable contention windows (Cat 4) for data.

Cat 4
Primary LBT Category for Data
02

Wideband Carrier Operation

NR-U supports carrier bandwidths up to 100 MHz in the 5 GHz band and wider in the 6 GHz band, far exceeding Wi-Fi's typical 20/40/80 MHz channels. To coexist with narrower-band incumbents, NR-U employs bandwidth part (BWP) switching and LBT sub-bands. A wideband carrier is divided into 20 MHz sub-bands, each performing independent LBT. Transmission occurs only on sub-bands sensed as free, maximizing throughput while respecting occupied channels.

100 MHz
Maximum Carrier Bandwidth
03

Anchor-Carrier and Standalone Modes

NR-U operates in two deployment architectures:

  • Non-Standalone (NSA): NR-U acts as a secondary cell anchored to a licensed LTE or NR primary cell. The licensed anchor carries control signaling and provides a reliable fallback.
  • Standalone (SA): NR-U operates entirely in unlicensed spectrum with no licensed anchor. This requires NR-U to handle all synchronization, paging, and random access procedures over unlicensed channels, enabling private networks in venues without owned spectrum.
Rel-16
NSA NR-U Introduced
Rel-17
SA NR-U Introduced
04

Channel Access Priority Classes

NR-U defines four Channel Access Priority Classes (CAPCs) to differentiate quality of service. Each CAPC maps to specific LBT parameters—defer duration and contention window size—that determine how aggressively a device contends for the channel.

  • CAPC 1: Lowest priority, longest backoff (best-effort data)
  • CAPC 4: Highest priority, shortest backoff (URLLC, control signaling) This allows NR-U to prioritize latency-sensitive traffic like industrial automation commands over background file transfers.
4
Priority Classes Defined
05

Discontinuous Transmission with COT Sharing

Once a device acquires the channel via LBT, it can transmit for a maximum Channel Occupancy Time (COT)—typically 8-10 ms. NR-U introduces COT sharing, where the initiating device (gNB or UE) can share its acquired COT with the responding device for uplink or downlink transmission without requiring a new LBT cycle. This reduces overhead and improves spectral efficiency. A COT structure indicator in the control channel signals the remaining transmission duration to all devices.

8-10 ms
Maximum COT Duration
06

Autonomous Uplink (AUL) Transmission

To reduce latency and signaling overhead in unlicensed spectrum, NR-U supports Configured Grant transmissions. The gNB pre-allocates periodic time-frequency resources to UEs, which can transmit uplink data immediately upon acquiring the channel via LBT—without waiting for a dynamic scheduling grant. This is critical for URLLC use cases in factory automation where every millisecond counts. The gNB blindly decodes these transmissions using the pre-configured parameters.

< 1 ms
Target URLLC Latency
UNLICENSED SPECTRUM TECHNOLOGY COMPARISON

NR-U vs. LAA vs. Wi-Fi 6E

Technical comparison of the three primary technologies operating in the 5 GHz and 6 GHz unlicensed bands, highlighting coexistence mechanisms, spectral efficiency, and deployment architectures.

FeatureNR-ULAA (LTE)Wi-Fi 6E

3GPP Release

Release 16

Release 13

Core Air Interface

5G NR

LTE

802.11ax

Primary Spectrum Bands

5 GHz, 6 GHz

5 GHz

2.4 GHz, 5 GHz, 6 GHz

Channel Access Mechanism

Category 4 LBT with COT sharing

Category 4 LBT

EDCA with CSMA/CA

Maximum Channel Bandwidth

100 MHz

20 MHz

160 MHz

Subcarrier Spacing

15/30/60 kHz

15 kHz

78.125 kHz

OFDMA Uplink Scheduling

Grant-based (gNB controlled)

Grant-based (eNB controlled)

Trigger-based (AP controlled)

Hybrid ARQ Process

Asynchronous, up to 16 processes

Synchronous, up to 8 processes

Standalone Operation

Anchor in Licensed Spectrum

Coexistence with Wi-Fi

LBT with adjustable contention window

LBT with fixed parameters

Native CSMA/CA

MIMO Layers (Max)

8

4

8

Modulation (Max DL)

256-QAM

256-QAM

1024-QAM

Latency Target

< 1 ms (URLLC capable)

2-4 ms

< 1 ms (OFDMA)

Synchronization Requirement

Loose (asynchronous capable)

Strict (synchronous)

Loose (asynchronous)

Acquisition Threshold

-72 dBm (adaptive)

-72 dBm (fixed)

-62 to -82 dBm (adaptive)

NR-U EXPLAINED

Frequently Asked Questions

Clear, technical answers to the most common questions about 5G New Radio operation in unlicensed spectrum and its coexistence mechanisms.

New Radio Unlicensed (NR-U) is a 3GPP standardized technology introduced in Release 16 that enables 5G New Radio (NR) operation in unlicensed and shared spectrum bands, such as the 5 GHz and 6 GHz bands. It works by aggregating unlicensed carriers with licensed anchors via carrier aggregation or operating in a standalone mode, while employing a Listen-Before-Talk (LBT) mechanism to ensure fair coexistence with Wi-Fi and other incumbents. Unlike Wi-Fi's purely asynchronous access, NR-U introduces a Channel Occupancy Time (COT) structure that allows a gNB to share its acquired channel time with user equipment, improving spectral efficiency. The technology supports both downlink and uplink transmissions, with flexible frame structures that adapt to the bursty nature of unlicensed channel access.

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.