mmWave & 5G

Conditional Handover

Q: What is the difference between conditional handover and baseline NR handover?

In baseline NR handover, the source gNB issues the RRCReconfiguration with the handover command only after it receives a measurement report and completes target preparation, so the command travels to the UE at the moment link quality is already degrading. In conditional handover (CHO), the network sends the prepared target configuration plus an execution condition while the radio link is still strong, and the UE stores it and applies it later, the instant a configured condition (for example RSRP of a candidate beam exceeding the serving beam by an offset for a Time-to-Trigger of 40 to 320 ms) is met. Decoupling preparation from execution removes the late-command race condition and cuts handover failures, especially at cell edges and in fast-fading mmWave deployments.

Q: Which CHO execution conditions are defined in 3GPP, and how many candidate cells can be prepared?

3GPP Release 16 defines two CHO trigger events: CondEvent A3 (a neighbor cell becomes better than the special cell by an offset) and CondEvent A5 (the serving cell falls below threshold1 while a neighbor rises above threshold2). Each candidate may carry up to two conditions that must both hold. The UE can be configured with up to 8 candidate target cells simultaneously in Release 16, each with its own pre-generated RRCReconfiguration. Release 17 extended the framework to conditional PSCell addition and change (CPAC) for dual connectivity, reusing the same prepare-then-execute logic for the secondary node.

Q: Why does conditional handover matter specifically for mmWave FR2 deployments?

At FR2 frequencies (24.25 to 52.6 GHz), the channel is highly directional and subject to rapid blockage, so a beam or cell can fade by 10 to 20 dB within a few milliseconds when a vehicle, person, or hand obstructs the path. A conventional handover command issued after the link has already collapsed often never reaches the UE, causing a radio link failure and a costly RRC re-establishment of 50 to 100 ms or more. Because CHO delivers the target configuration in advance, the UE can switch the moment a stronger beam crosses the threshold without any further downlink signaling, shrinking the interruption to roughly the random-access time on the target cell and dramatically improving mobility robustness for high-band 5G.

/kuhn-DISH-uh-nuhl HAND-oh-ver/ (CHO)

Standardized in 3GPP Release 16, this 5G NR mobility procedure separates handover preparation from handover execution. The serving gNB pre-configures up to eight candidate target cells, each bundled with an execution condition, and hands the prepared handover command to the UE while the radio link is still healthy. The UE stores the configuration and triggers the switch itself the moment a configured radio metric is met, for example a candidate cell's RSRP exceeding the serving cell by a 2 to 6 dB offset held for a Time-to-Trigger of 40 to 320 ms. By removing the late-command race condition that plagues conventional handover, CHO sharply reduces mobility failures at cell edges and in fast-fading beamforming FR2 deployments where a beam can fade 10 to 20 dB within milliseconds.

Category: mmWave & 5G

Standard: 3GPP Rel-16 (TS 38.331)

Max candidates: 8 target cells

How Prepare-Then-Execute Mobility Works

Conventional NR handover is reactive: the UE sends a measurement report, the source gNB negotiates resources with the target over the Xn interface, and only then does it transmit the RRCReconfiguration carrying the handover command. That command must reach the UE over a radio link that, by definition, is already weakening, which is exactly when the downlink is least reliable. Conditional handover inverts the timeline. The network prepares candidate targets early, while signal-to-interference-plus-noise ratio is still high, and ships each candidate's RRCReconfiguration together with one or two execution conditions. The UE caches these configurations and evaluates the conditions locally against its ongoing measurements.

Two trigger events are defined for CHO. CondEvent A3 fires when a neighbor cell becomes better than the special cell by a configured offset, and CondEvent A5 fires when the serving cell drops below threshold1 while a neighbor rises above threshold2. A candidate may be guarded by up to two conditions that must both be satisfied, which lets operators tune aggressiveness against ping-pong risk. When a condition holds for the Time-to-Trigger, the UE detaches from the source, performs random access on the chosen target, and sends RRCReconfigurationComplete; no further downlink command is needed, so the interruption shrinks to roughly the random-access duration on the target cell.

Because the prepared configurations consume target resources that may never be used, the network balances the number of candidates against admission-control overhead. Release 17 generalized the same logic to conditional PSCell addition and change (CPAC), extending robust prepare-then-execute mobility to the secondary node in dual connectivity arrangements.

CHO Execution-Condition Equations

CondEvent A3 entering condition (neighbor better than SpCell):
M_n + Ofn + Ocn − Hys > M_p + Ofp + Ocp + Off

CondEvent A5 entering condition (SpCell below T1, neighbor above T2):
M_p + Hys < Thresh1 AND M_n + Ofn + Ocn − Hys > Thresh2

Interruption-time approximation:
T_int ≈ T_TTT + T_RACH; with T_TTT = 40 to 320 ms, T_RACH ≈ 5 to 15 ms

M_n, M_p = measured RSRP/RSRQ of neighbor and serving (SpCell); Ofn/Ofp = frequency offsets; Ocn/Ocp = cell-individual offsets; Hys = hysteresis (0 to 15 dB); Off = A3 offset; Thresh1/Thresh2 = A5 thresholds; T_TTT = Time-to-Trigger. Each candidate may bind up to two such conditions.

Mobility Mechanism Comparison

Mechanism	Trigger / executor	Command timing	Typical interruption	Failure rate (cell edge)	3GPP origin
Conditional handover (CHO)	UE, CondEvent A3/A5	Prepared early, executed on condition	~5 to 20 ms	Very low	Rel-16 (TS 38.331)
Baseline NR handover	gNB, after meas. report	At link degradation	~20 to 50 ms	Moderate to high	Rel-15
DAPS handover	gNB, dual active protocol	Make-before-break	~0 ms (data)	Low	Rel-16
Conditional PSCell change (CPAC)	UE, secondary node	Prepared early, on condition	~5 to 20 ms	Very low	Rel-17
Beam failure recovery	UE, intra-cell	On beam failure detection	~10 to 30 ms	n/a (intra-cell)	Rel-15

Common Questions

Frequently Asked Questions

What is the difference between conditional handover and baseline NR handover?

Baseline handover sends the RRCReconfiguration command only after the source gNB receives a measurement report and prepares the target, so it arrives while the link is already degrading. CHO ships the prepared target configuration plus an execution condition early, while the link is strong; the UE stores it and applies it itself the instant a configured condition (such as a candidate RSRP exceeding the serving cell by an offset held for a 40 to 320 ms Time-to-Trigger) is met, removing the late-command race condition.

Which CHO execution conditions are defined in 3GPP, and how many candidate cells can be prepared?

Release 16 defines CondEvent A3 (neighbor better than the special cell by an offset) and CondEvent A5 (serving cell below threshold1 while a neighbor exceeds threshold2). A candidate may carry up to two conditions that must both hold, and the UE can be configured with up to 8 candidate target cells, each with its own pre-generated RRCReconfiguration. Release 17 extended the framework to conditional PSCell addition and change for dual connectivity.

Why does conditional handover matter specifically for mmWave FR2 deployments?

At FR2 (24.25 to 52.6 GHz) the channel is highly directional and prone to rapid blockage, so a beam can fade 10 to 20 dB within a few milliseconds. A conventional command issued after the link collapses may never reach the UE, forcing a 50 to 100 ms RRC re-establishment. Because CHO delivers the target configuration in advance, the UE switches the moment a stronger beam crosses the threshold with no further downlink signaling, shrinking the interruption to roughly the target random-access time.

Related Terms

← Concession Conditional MRI Safety →