Handover is the procedure by which a mobile network transfers an active connection from one cell, beam or radio access technology to another while the session continues, so that a moving user keeps a voice call or data flow without interruption. In telecommunications, and especially in cellular network design, handover (called handoff in North American usage) is the mechanism that turns a patchwork of individual cells into continuous network coverage.
The procedure combines radio measurements by the device, decisions by the network and signaling between base stations and the core network. Its quality is judged by how rarely it fails, how short the interruption is, and how few unnecessary handovers it triggers.
How a handover works step by step
A handover follows a fixed cycle of measurement, decision and execution. The device continuously measures the serving cell and neighboring cells, reports when configured conditions are met, and the network then prepares a target cell and commands the device to move. In LTE and 5G NR this is a network-controlled, device-assisted process: the user equipment supplies the measurements, but the base station makes the decision.
The key radio quantities are RSRP (reference signal received power, typically between -140 dBm and -44 dBm), RSRQ (reference signal received quality) and SINR (signal to interference plus noise ratio). The network configures measurement events that define when a report is sent. The following table summarizes the most common events defined by 3GPP for LTE and NR.
| Event | Trigger condition | Typical purpose |
|---|---|---|
| A1 | Serving cell becomes better than a threshold | Stop neighbor measurements |
| A2 | Serving cell becomes worse than a threshold | Start inter-frequency measurements |
| A3 | Neighbor becomes offset better than serving cell | Classic intra-frequency handover |
| A4 | Neighbor becomes better than a threshold | Load balancing to another layer |
| A5 | Serving below one threshold, neighbor above another | Coverage-driven inter-frequency move |
| B1 and B2 | Inter-RAT neighbor meets a threshold | Move to or from another technology |
Once the source base station decides, it sends a handover request to the target, which performs admission control, reserves resources and returns a configuration. The source forwards that configuration to the device as a handover command. The device then synchronizes with the target cell, performs random access and confirms completion, after which the core network switches the data path to the new cell. Data buffered in the source during the switch is forwarded over the X2 (LTE) or Xn (NR) interface to avoid packet loss.
Types of handover in cellular networks
Handovers are classified by whether the old link is released first, by which frequencies and technologies are involved, and by which network nodes take part. The categories overlap: a single event can be hard, inter-frequency and Xn-based at the same time.
| Type | Defining feature | Typical technology |
|---|---|---|
| Hard handover | Old link released before new link is used | GSM, LTE, 5G NR |
| Soft handover | Device connected to several cells at once | CDMA and UMTS (WCDMA) |
| Softer handover | Several sectors of the same site combined | UMTS |
| Intra-frequency | Source and target share a carrier | All cellular systems |
| Inter-frequency | Target on another frequency band or carrier | Multi-layer LTE and NR |
| Inter-RAT | Target uses a different radio technology | NR to LTE, LTE to UMTS |
| Conditional handover | Device executes when a preset condition is met | LTE Release 16, 5G NR |
Hard handover is often described as break-before-make, and the interruption in LTE is typically in the range of 30 to 60 milliseconds, short enough to be inaudible in a voice call. Soft handover is make-before-break and relies on macro-diversity combining, which suited CDMA because neighboring cells reuse the same frequency. Newer releases add dual active protocol stack (DAPS) handover, which keeps both links alive briefly to approach zero interruption for demanding services.
Handover parameters and how operators tune them
Handover behavior is governed by a small set of parameters that operators adjust per cell. The two most influential are hysteresis, a margin in decibels that the neighbor must exceed, and time-to-trigger (TTT), the period during which the condition must hold before a report is sent. Together they balance responsiveness against stability.
| Parameter | Typical range | Effect of a higher value |
|---|---|---|
| Hysteresis | 0 to 6 dB, often 2 to 3 dB | Fewer ping-pongs, later handovers |
| Time-to-trigger | 40 to 640 ms, often 160 to 320 ms | More stability, risk of late handover |
| Cell individual offset | -24 to +24 dB | Shifts the effective cell border |
| A3 offset | 0 to 5 dB | Neighbor must be clearly stronger |
| Filter coefficient | Layer 3 averaging factor | Smoother but slower measurements |
Tuning is rarely static. Self-organizing network (SON) functions such as mobility robustness optimization analyze failure statistics and adjust offsets automatically, while automatic neighbor relations discover missing neighbor cells from device reports. Without correct neighbor lists, even well-tuned parameters cannot prevent drops.
Handover compared with related mobility procedures
Handover applies only to devices in connected mode, which means devices with an active radio connection. Idle devices use cell reselection, which the device decides on its own without network signaling. The distinction matters because the two procedures use different parameters and fail in different ways.
| Procedure | Device state | Who decides | Signaling load |
|---|---|---|---|
| Handover | Connected | Network | High, per event |
| Cell reselection | Idle or inactive | Device | None until area change |
| Tracking area update | Idle | Device, reported to core | Low, periodic |
| Redirection | Connected, then released | Network | Moderate, with interruption |
| Secondary cell change | Connected with carrier aggregation | Network | Low, no anchor change |
Related concepts in the same field interact closely with handover. Carrier aggregation and dual connectivity can change a secondary carrier without a full handover of the primary cell, and a congested backhaul can delay data forwarding, which lengthens the effective interruption even when radio signaling is fast.
Failure modes and common misconceptions
Handover problems fall into a few recognizable patterns, and key performance indicators such as handover success rate (commonly expected above 98 to 99 percent) and the share of radio link failures reveal which one is present.
- Too-late handover: the signal collapses before the report arrives, typical at fast street corners and on highways.
- Too-early handover: the device moves to a target that cannot yet sustain the link and fails shortly afterward.
- Ping-pong handover: the device bounces between two cells of similar strength, wasting signaling and battery.
- Handover to wrong cell: a missing or incorrect neighbor relation sends the device to a weaker cell.
- Misconception: more handovers do not mean better service; unnecessary handovers increase failure risk and signaling load.
High-speed rail, dense small-cell deployments and millimeter-wave beams intensify these problems because the dwell time in each cell shrinks to seconds. This is the main reason why 5G introduced conditional handover and beam-level mobility, which prepare targets in advance and let the device execute at the right moment.
A plain-language summary
A mobile network is made of many small radio zones called cells, and each one only reaches so far. When you walk or drive while on a call, your phone keeps leaving one zone and entering another. Handover is the quiet relay that passes your call from one cell to the next, much like runners passing a baton without stopping.
Your phone constantly checks how well it hears the nearby cells and tells the network. The network then picks the best next cell, gets it ready and tells your phone to switch. When it works well, you never notice it; when it fails, you hear a dropped call or see a stalled video.
An example from practice
A regional operator received repeated complaints about dropped calls on a highway interchange where traffic moves at about 110 km/h. Drive tests showed that devices left the coverage of one LTE cell so quickly that the A3 report arrived too late, with a time-to-trigger of 320 ms and a hysteresis of 3 dB.
The radio optimization team reduced time-to-trigger to 160 ms for the highway-facing cells, added a missing neighbor relation to a cell on the opposite side of the interchange, and applied a cell individual offset of 2 dB so the target was recognized earlier. Within two weeks the handover success rate on the affected cells rose from about 96 percent to above 99 percent, while ping-pong events in the nearby town center stayed unchanged because the new settings were applied only to the highway sectors.