Mobility in 5G NR

Handover (HO) is the procedure that transfers a UE's connection from a source cell to a target cell without interrupting active services. 5G NR defines multiple handover types to handle different network topologies and inter-system scenarios. The baseline handover procedures are specified in TS 38.300 Section 9.2.3 for intra-NR mobility and TS 38.331 Section 5.3.5 for the RRC layer.

The three primary handover types in 5G NR are:

• Xn handover: Direct handover between gNBs connected via Xn interface (most common)
• N2 handover: Handover via the AMF when no Xn interface exists between source and target
• Inter-RAT handover: Handover between 5G NR and LTE (or other RATs)

Measurement Events

Before any handover can be triggered, the UE must perform measurements and report results to the serving gNB. NR defines measurement events in TS 38.331 Section 5.5.4 that control when the UE sends a Measurement Report.

The Time-to-Trigger (TTT) prevents ping-pong handovers by requiring the trigger condition to be continuously satisfied for the specified duration. The hysteresis adds a measurement offset to prevent rapid switching between cells with similar signal levels.

Event A3 is the workhorse of intra-frequency mobility. The UE triggers a report when: `

Mn + Ofn + Ocn - Hys > Ms + Ofs + Ocs + Off

`

Where Mn is the neighbor measurement, Ms is the serving measurement, Off is the A3 offset, and Hys is the hysteresis.

Xn Handover: Step-by-Step Message Flow

The Xn-based handover is the most common type and the fastest because signaling goes directly between gNBs over the Xn interface without involving the core network in the path switch until after the UE has moved. Defined in TS 38.300 Section 9.2.3.2.1.

1. UE sends Measurement Report --- UE reports Event A3 triggered to source gNB via RRC
2. Source gNB makes HO decision --- Source evaluates the report, selects target cell, and decides to initiate handover
3. HANDOVER REQUEST (Xn-AP) --- Source gNB sends Xn-AP Handover Request to target gNB, including UE context (security, bearers, UE capabilities)
4. Target gNB performs admission control --- Target checks resource availability and QoS requirements
5. HANDOVER REQUEST ACKNOWLEDGE (Xn-AP) --- Target gNB responds with ACK, including the transparent container with target cell configuration
6. RRCReconfiguration (HO Command) --- Source gNB forwards the target cell configuration to UE via RRCReconfiguration with reconfigurationWithSync
7. UE detaches from source, syncs to target --- UE performs random access on target cell using dedicated RACH preamble
8. RRCReconfigurationComplete --- UE confirms successful reconfiguration to target gNB
9. PATH SWITCH REQUEST (N2) --- Target gNB sends N2 Path Switch Request to AMF to update the CN path
10. UPF path update --- AMF instructs SMF/UPF to switch the downlink tunnel endpoint to target gNB
11. PATH SWITCH REQUEST ACKNOWLEDGE --- AMF confirms path switch to target gNB
12. UE CONTEXT RELEASE (Xn-AP) --- Target gNB sends release to source gNB to free old UE context and forwarded data

The handover interruption time (HIT) for Xn handover is typically 20--40 ms in well-optimized networks, dominated by the random access procedure on the target cell.

N2 Handover: When Xn Is Unavailable

When source and target gNBs have no direct Xn interface (different vendors, different administrative domains), the handover is routed through the AMF via N2. Defined in TS 23.502 Section 4.9.1.3.2.

The N2 handover adds approximately 20--40 ms of additional latency due to the extra signaling hop through the AMF. This makes Xn handover strongly preferred for intra-frequency mobility where possible.

Inter-RAT Handover: 5G to LTE Fallback

When 5G NR coverage is lost, the UE falls back to LTE via inter-RAT handover. This is critical for ensuring service continuity at 5G coverage edges. The procedure follows TS 36.331 Section 5.4.3 on the LTE side and TS 38.331 Section 5.4.3 on the NR side.

The inter-RAT handover from NR to LTE proceeds as follows:

1. UE reports Event B2 (NR serving below threshold AND LTE neighbor above threshold)
2. Source gNB initiates handover via AMF (always N2-based since the target is a different RAT)
3. AMF communicates with MME (via N26 interface if available) or directly with target eNB
4. Target eNB performs admission control and returns HO command
5. Source gNB sends MobilityFromNRCommand to UE with target LTE configuration
6. UE switches to LTE, performs RACH on target eNB
7. UE sends RRCConnectionReconfigurationComplete to target eNB

If the N26 interface between AMF and MME is not deployed (increasingly common in SA deployments), the UE must perform a full registration on the LTE side, increasing the interruption time to 200--500 ms.

Handover Timers

NR defines several timers that control handover failure detection and recovery, specified in TS 38.331 Section 5.3.5.4.

T304 expiry is the most critical handover failure scenario. If the UE cannot complete random access on the target cell within T304, it declares a handover failure and initiates the RRC re-establishment procedure on the best available cell. If re-establishment also fails (T311 expires), the UE enters RRC_IDLE and must perform a new connection setup.

Worked Example: T304 Timeout Calculation

In a dense urban network with average RACH load, calculate whether T304 = 100 ms is sufficient:

`

RACH preamble transmission: 1 ms (1 RACH occasion)

RACH Response Window (ra-ResponseWindow): 10 ms

MSG3 (RRC Reconfiguration Complete): 1 ms

Contention resolution: 8 ms (worst case with CBRA)

Processing delays: 5 ms

───────────────────────────────────────

Total RACH procedure: 25 ms

`

With dedicated preamble (CFRA, no contention resolution), total drops to approximately 15 ms. T304 = 100 ms provides a 4--6x safety margin. Networks with high RACH load or FR2 beam sweeping may increase T304 to 200 ms.

Worked Example: HO Ping-Pong Avoidance

Calculate the minimum A3 offset to avoid ping-pong at a cell edge where both cells show RSRP fluctuations of +/- 4 dB:

`

Required condition: Offset > 2 * (Signal fluctuation - Hysteresis)

Offset > 2 * (4 - 3) = 2 dB

With TTT = 480 ms, the UE must sustain this condition

for 480 ms, further reducing ping-pong probability.

Recommended A3 offset: 3 dB (provides additional margin)

`

Real Operator Handover Performance

SK Telecom's 99.1% HO success rate is among the highest reported globally, attributed to dense cell grid (350 m ISD in Seoul), single-vendor Samsung RAN, and aggressive parameter optimization including A3 offset of 2 dB and TTT of 320 ms.

Rakuten Mobile's slightly lower HO success rate (96.8%) and higher HIT (42 ms) reflect the challenges of multi-vendor O-RAN deployments where Xn interoperability between different RAN vendors introduces additional processing delay.

Handover Optimization Parameters

Network engineers tune these parameters to balance HO success rate against ping-pong rate:

• A3 Offset: Lower values (1--2 dB) trigger earlier HO, improving success rate but increasing ping-pong. Higher values (4--6 dB) delay HO, reducing ping-pong but risking late HO failures.
• TTT: Shorter TTT (128--256 ms) for high-speed mobility (highways, rail). Longer TTT (480--1024 ms) for pedestrian/stationary scenarios.
• Hysteresis: Typically 1--3 dB. Higher values stabilize measurements but reduce HO responsiveness.
• Layer-3 filter coefficient (k): Controls RSRP/RSRQ smoothing. k=4 (default) provides moderate filtering. k=1 for fast mobility, k=6 for stationary scenarios.

Key Takeaway: Xn handover is the preferred 5G mobility mechanism with 20--40 ms interruption time and 97--99% success rates in mature networks. Event A3 with properly tuned offset and TTT is the primary intra-frequency trigger. For inter-RAT fallback to LTE, deploy the N26 interface between AMF and MME to avoid full re-registration delays of 200+ ms.