What are the beam management procedures in 5G NR?

5G NR defines four beam management procedures. P1 (initial acquisition) uses SSB beam sweeping where the gNB transmits SSBs in up to 64 directions at mmWave (8 at sub-6 GHz) within a burst period of 5 ms. The UE measures L1-RSRP on each SSB index and reports the best beam(s). P2 (gNB transmit beam refinement) uses CSI-RS resources beamformed within the angular neighborhood of the best SSB beam, using narrower beams for finer angular resolution. P3 (UE receive beam refinement) keeps the gNB transmit beam fixed while the UE sweeps its own receive beam to find the optimal direction. The result is an optimized beam pair link (gNB TX beam plus UE RX beam) that maximizes the received signal quality. Beam failure recovery triggers when L1-RSRP drops below a configured threshold for a consecutive number of measurement instances, prompting the UE to identify a new candidate beam and initiate a RACH-based recovery procedure.

5G NR Procedures

Beam Management

Q: How fast can 5G NR switch beams?

Beam switching speed depends on the procedure. L1 beam indication via DCI (Downlink Control Information) can switch the serving beam within a single slot, which is 0.5 ms at 30 kHz SCS or 0.125 ms at 120 kHz SCS. This is used for scheduled beam switching when the network knows the best beam from recent reports. Beam failure recovery takes longer: the UE must detect beam failure (typically requiring consecutive RSRP measurements below threshold over 2 to 10 ms), select a candidate beam, transmit a RACH preamble on the candidate beam's PRACH occasion, and wait for the gNB response. Total beam failure recovery latency is 10 to 50 ms depending on configuration and PRACH periodicity. For connected-mode beam tracking without failure, MAC CE-based beam switching completes in 3 to 5 ms.

Q: What is beam correspondence?

Beam correspondence is the property where a device's best transmit beam direction is the same as its best receive beam direction. When beam correspondence holds, the UE can determine its optimal uplink transmit beam from downlink measurements alone, without requiring separate uplink beam sweeping. This significantly reduces beam management overhead. At sub-6 GHz, beam correspondence generally holds because antenna patterns are similar for TX and RX. At mmWave, beam correspondence may not hold due to different TX and RX antenna array configurations, different front-end architectures (separate PA and LNA modules), or frequency-dependent beam squinting in TDD systems where UL and DL use different frequencies in FDD. When beam correspondence does not hold, the gNB must configure separate uplink beam management procedures using SRS (Sounding Reference Signal) resources for the UE to sweep its transmit beams.

/beem man-ij-ment/

The 5G NR Layer 1/Layer 2 framework for acquiring, tracking, and recovering directional beam pairs between gNB and UE. Four procedures: P1 initial acquisition (SSB sweeping, up to 64 beams at mmWave), P2 gNB beam refinement (CSI-RS), P3 UE beam refinement, and beam failure recovery (RACH on candidate beams). L1-RSRP measurement drives all beam decisions. Beam switch latency: 0.5 ms (DCI-based) to 50 ms (failure recovery). Essential for mmWave 5G where narrow beams require continuous tracking.

Beams: Up to 64 (mmWave)

Switch: 0.5-50 ms

Metric: L1-RSRP

Understanding Beam Management

At mmWave frequencies, both the base station and user equipment use narrow beams (5-30°) to overcome the high path loss. This creates a fundamental challenge: the gNB and UE must find each other's beams before communication can begin, and they must continuously track the optimal beam pair as the user moves. Beam management is the set of 3GPP-defined procedures that solve this problem. It is the "spatial dimension" of the 5G NR air interface, analogous to how frequency synchronization solves the spectral dimension.

The framework operates hierarchically. Coarse acquisition (P1) sweeps wide beams across all directions to establish initial contact. Fine refinement (P2, P3) narrows the beam using CSI-RS and UE-side sweeping. Continuous monitoring tracks beam quality, and failure recovery provides a safety net when the serving beam is lost due to blockage, rotation, or movement. The entire system must operate fast enough that users perceive seamless connectivity even while walking or riding in vehicles.

Beam Management Timing

SSB beam sweep period:
T_SSB = 5, 10, 20, 40, 80, 160 ms
Default: 20 ms (sub-6), 20 ms (mmWave)
Max SSB beams: 4 (<3 GHz), 8 (<6 GHz),
64 (6-52.6 GHz)

Beam failure detection:
BFD timer: 2-10 ms (configurable)
Threshold: RSRP < Q_out for N instances
Q_out = -156 + offset dBm/SCS

Beam switch latency:
DCI-based (L1): 1 slot = 0.5 ms @ 30 kHz
MAC CE-based (L2): 3-5 ms
Failure recovery: 10-50 ms

Beam tracking rate:
CSI-RS: up to every 4 ms
Required for >30 km/h at mmWave

Beam Management Procedure Comparison

Procedure	Signal	Scope	Resolution	Latency	Trigger
P1 (Acquisition)	SSB	All directions	Coarse (wide)	5-20 ms	Initial access
P2 (gNB refine)	CSI-RS	Best SSB region	Fine (narrow)	1-5 ms	After P1
P3 (UE refine)	CSI-RS	UE RX beams	Fine	1-5 ms	After P2
Tracking	CSI-RS/SSB	Serving + neighbors	Continuous	4-20 ms	Periodic
Failure Recovery	RACH	Candidate beams	Coarse	10-50 ms	RSRP < Q_out

Common Questions

Frequently Asked Questions

What are the beam management procedures?

P1: SSB beam sweeping across all directions (up to 64 at mmWave) for initial acquisition. P2: CSI-RS refinement within the best SSB region using narrower beams. P3: UE sweeps its receive beams with fixed gNB transmit beam. Beam failure recovery: UE detects RSRP below threshold, selects candidate beam from last sweep, initiates RACH recovery. Total framework enables optimized beam pair link between gNB TX and UE RX.

How fast can 5G NR switch beams?

DCI-based (L1): single slot, 0.5 ms at 30 kHz SCS, 0.125 ms at 120 kHz SCS. MAC CE-based (L2): 3-5 ms. Beam failure recovery: 10-50 ms total (detection + candidate selection + RACH + response). For connected-mode tracking without failure, beam updates at CSI-RS rate of every 4 ms handle mobility up to vehicular speeds at mmWave.

What is beam correspondence?

The property where a device's best TX beam equals its best RX beam direction. When it holds (common at sub-6 GHz), the UE derives its uplink beam from downlink measurements alone, eliminating UL beam sweeping overhead. At mmWave, correspondence may fail due to different TX/RX array configurations or front-end architectures. Without correspondence, separate SRS-based uplink beam management is needed.

Related Terms

← Beam Hopping Beam Port →