When does 2-Step RACH fall back to 4-Step?

The gNB configures both 2-Step and 4-Step RACH resources. The UE selects 2-Step when channel conditions are good (path loss below a configured threshold, estimated SINR above a minimum). If MsgA transmission fails after a configured number of retries, the UE falls back to 4-Step RACH, which is more robust because the gNB can adjust timing advance and power control before the UE sends data. Fallback also occurs when the UE is at cell edge or in poor coverage where the combined preamble+data transmission in MsgA is unlikely to succeed.

What is the structure of MsgA?

MsgA consists of two parts transmitted in the same slot or adjacent slots: (1) PRACH preamble: identical to the 4-Step Msg1 preamble (short or long format), used for timing estimation and UE identification. (2) PUSCH payload: carries the UE's RRC connection request, BSR, and optionally small uplink data. The PUSCH uses a configured MCS (typically QPSK rate 1/2 for robustness) and is mapped to resources associated with the preamble index. The gNB must decode both preamble and PUSCH to process MsgA. If only the preamble is detected (PUSCH decoding fails), the gNB can fall back to 4-Step by sending a standard RAR.

mmWave & 5G

2-Step RACH

Q: How does 2-Step RACH reduce latency?

4-Step RACH requires four over-the-air exchanges: Msg1 (preamble), Msg2 (RAR), Msg3 (RRC request), Msg4 (contention resolution). Each exchange takes at least one slot (0.5 ms at 30 kHz SCS) plus processing time, totaling 10-15 ms. 2-Step RACH combines Msg1+Msg3 into MsgA and Msg2+Msg4 into MsgB, cutting the procedure to two exchanges (approximately 5-8 ms). This is critical for URLLC applications targeting 1 ms latency, handover optimization, and beam recovery in mmWave where link interruptions must be minimized.

A simplified random access channel procedure introduced in 3GPP Release 16 for 5G NR that reduces the legacy 4-Step RACH to two over-the-air exchanges. The UE combines the preamble and uplink data into a single transmission (MsgA), and the gNB responds with contention resolution, timing advance, and C-RNTI assignment in a single response (MsgB). This cuts initial access latency by approximately 50%, from 10-15 ms to 5-8 ms, which is critical for URLLC applications and mmWave beam recovery.

Category: mmWave & 5G

3GPP Release: Release 16

Latency: ~5-8 ms (vs. 10-15 ms 4-Step)

Understanding 2-Step RACH

Random access is how a UE establishes initial connection with a cell, performs handover, or recovers from beam failure. The legacy 4-Step procedure (Msg1-Msg4) was inherited from LTE and works reliably but requires four round-trips. In scenarios demanding ultra-low latency (factory automation, autonomous vehicles, AR/VR) or rapid beam recovery (mmWave), the 10-15 ms overhead is too high.

2-Step RACH compresses the procedure by having the UE transmit its data payload alongside the preamble. This is a calculated risk: without prior timing advance, the PUSCH portion of MsgA has reduced reliability. The gNB mitigates this by allocating wider timing tolerances in the PUSCH resources and using robust MCS (QPSK rate 1/2). If the PUSCH fails, the system gracefully falls back to the 4-Step procedure.

2-Step vs. 4-Step RACH

4-Step RACH (legacy):
Msg1 (UE→gNB): PRACH preamble
Msg2 (gNB→UE): RAR (TA, UL grant, TC-RNTI)
Msg3 (UE→gNB): RRC request on PUSCH
Msg4 (gNB→UE): Contention resolution
Total: ~10-15 ms

2-Step RACH (Rel-16):
MsgA (UE→gNB): PRACH preamble + PUSCH payload
MsgB (gNB→UE): TA + C-RNTI + contention resolution
Total: ~5-8 ms

Fallback: If MsgA PUSCH fails, gNB sends RAR → UE continues with 4-Step Msg3/Msg4.

RACH Procedure Comparison

Property	4-Step RACH	2-Step RACH
Messages	4 (Msg1-Msg4)	2 (MsgA, MsgB)
Latency	10-15 ms	5-8 ms
Reliability	Higher (TA before data)	Lower (no prior TA)
Coverage	Cell-edge robust	Requires good SINR
3GPP Release	Release 15	Release 16
Use case	General access	URLLC, beam recovery

Common Questions

Frequently Asked Questions

How does 2-Step RACH reduce latency?

Combines Msg1+Msg3 into MsgA and Msg2+Msg4 into MsgB, cutting four exchanges to two. Saves ~5-7 ms. Critical for URLLC (1 ms target), handover, and mmWave beam recovery where link interruptions must be minimized.

When does it fall back to 4-Step?

When channel conditions are poor (high path loss, low SINR) or after MsgA retries fail. The gNB can also force fallback by responding with a standard RAR instead of MsgB. Cell-edge UEs default to 4-Step for reliability.

What is MsgA structure?

PRACH preamble (timing estimation) + PUSCH payload (RRC request, BSR, optional small data). PUSCH uses robust MCS (QPSK R=1/2). If gNB decodes preamble but not PUSCH, it falls back to 4-Step gracefully.

Related Terms

← 16-QAM 32T32R →