Ferkans — Interactive Telecom Tutor

Seamless Mobility in Cellular Networks

A defining feature of cellular systems is that users move between cells without dropping their connections. The handover (or handoff) procedure transfers a user's radio resources from a source cell to a target cell. Getting this right is a delicate balance: trigger the handover too early and the user ping-pongs back and forth between cells, wasting signalling resources; trigger it too late and the user drops the connection due to poor signal quality. The key parameters — hysteresis margin and time-to-trigger — must be tuned to the user's velocity and the propagation environment. 3GPP defines precise measurement events (A1--A6) that govern when handover is initiated, with the A3 event being the most fundamental.

Definition:
A3 Event and Handover Trigger

The A3 event in 3GPP (LTE/NR) is the standard intra-frequency handover trigger. It fires when the neighbouring cell's reference signal received power (RSRP) exceeds the serving cell's RSRP by a hysteresis margin for a sustained duration:

$\text{RSRP}_{\text{target}} > \text{RSRP}_{\text{source}} + H_{\text{hys}}$

for a continuous duration of $T_{\text{TTT}}$ (time-to-trigger).

The handover parameters are:

Hysteresis margin $H_{\text{hys}}$ (dB): prevents handover due to small signal fluctuations. Typical range: 1--6 dB.
Time-to-trigger $T_{\text{TTT}}$ (ms): the A3 condition must hold continuously for this duration before the handover is initiated. Typical values: 40--640 ms.
Layer-3 filtering coefficient $k$ : smooths the RSRP measurements to reduce noise-induced triggers.

Other 3GPP events include A1 (serving becomes better than threshold), A2 (serving becomes worse than threshold), A4 (neighbour becomes better than threshold), A5 (serving worse and neighbour better than respective thresholds), and A6 (neighbour becomes offset better than secondary cell).

Definition:
Ping-Pong Handover

A ping-pong handover occurs when a user repeatedly hands over between two cells within a short time interval. Formally, a handover from cell $A$ to cell $B$ followed by a return handover from $B$ to $A$ within a time window $T_{\text{pp}}$ (typically 1--5 seconds) is classified as a ping-pong.

The ping-pong rate $R_{\text{pp}}$ is the fraction of handovers that are ping-pong:

$R_{\text{pp}} = \frac{\text{number of ping-pong handovers}} {\text{total number of handovers}}$

Ping-pong rate increases with UE velocity (more frequent cell boundary crossings), decreases with larger hysteresis margin, and decreases with longer TTT.

Theorem: Handover Rate and Ping-Pong Trade-Off

Consider a UE moving at velocity $v$ through a cellular network with average cell radius $R$ . Under a simplified model with Poisson-distributed cell boundary crossings, the handover rate is:

$\mu_{\text{HO}} = \frac{2v}{\pi R} \quad\text{(handovers per second)}$

The handover failure probability $p_{\text{fail}}$ increases with velocity because the handover preparation time $T_{\text{prep}} = T_{\text{TTT}} + T_{\text{exec}}$ may exceed the time the UE spends in the overlapping coverage zone:

$p_{\text{fail}} \approx \Pr\!\left[T_{\text{prep}} > \frac{W_{\text{overlap}}}{v}\right]$

where $W_{\text{overlap}}$ is the width of the handover region. The ping-pong probability increases as:

$p_{\text{pp}} \propto \exp\!\left(-\frac{H_{\text{hys}}^2}{2\sigma_{\text{sf}}^2}\right) \cdot \exp\!\left(-\frac{v \cdot T_{\text{TTT}}}{R}\right)$

showing that both hysteresis and TTT suppress ping-pong, but at the cost of increased handover delay and potential failures.

The handover rate scales linearly with velocity $v$ — a UE at 120 km/h crosses cell boundaries 4 times more often than at 30 km/h. The hysteresis margin prevents handover unless the target cell is definitively stronger, while the TTT ensures this condition is not transient. However, excessive hysteresis or TTT delays the handover so much that the UE leaves the source cell's coverage before the handover completes, causing a radio link failure (RLF).

Proof

Handover rate from geometric probability

For a UE moving in a random direction at speed $v$ through a PPP network with intensity $\lambda$ , the rate of crossing Voronoi cell boundaries is (from stochastic geometry):

$\mu_{\text{HO}} = \frac{4v\sqrt{\lambda}}{\pi}$

With $\lambda = 1/(\pi R^2)$ (relating density to cell radius):

$\mu_{\text{HO}} = \frac{4v}{\pi\sqrt{\pi}R} \approx \frac{2v}{\pi R}$

Ping-pong analysis

A ping-pong occurs when shadow fading causes a transient A3 event. The probability that a shadow-fading fluctuation exceeds $H_{\text{hys}}$ is $Q(H_{\text{hys}}/\sigma_{\text{sf}})$ . The TTT provides additional filtering: the fluctuation must persist for duration $T_{\text{TTT}}$ , which requires the UE to remain in the transition zone. Combining:

$p_{\text{pp}} \propto Q\!\left(\frac{H_{\text{hys}}}{\sigma_{\text{sf}}}\right) \cdot \exp\!\left(-\frac{v T_{\text{TTT}}}{R}\right)$

showing the joint effect of hysteresis and TTT.

Optimal parameter design

The optimal $H_{\text{hys}}$ and $T_{\text{TTT}}$ minimise a cost function balancing ping-pong rate and handover failure:

$J = w_1 \cdot p_{\text{pp}} + w_2 \cdot p_{\text{fail}}$

This is typically solved via mobility robustness optimisation (MRO), a self-organising network (SON) function in LTE/NR. $\blacksquare$

A3 Handover Event with Hysteresis and TTT

Animated RSRP traces from source and target cells as a UE moves between them. Shows the hysteresis margin, the time-to-trigger window, and the resulting handover decision.

The A3 event triggers when the target RSRP exceeds the source by

H_{\mathrm{hys}}

for a sustained duration

T_{\mathrm{TTT}}

.

Handover Analysis

Explore the trade-off between handover robustness and ping-pong rate. Adjust the hysteresis margin, time-to-trigger, and UE velocity to observe their effects on the handover success rate, ping-pong rate, and handover delay. Higher velocity requires faster handover execution but also increases the risk of ping-pong due to more frequent cell boundary crossings. The simulation shows the RSRP traces from two cells and the resulting handover decisions.

Parameters

Hysteresis

H_{\mathrm{hys}}

(dB)3

Time-to-trigger

T_{\mathrm{TTT}}

(ms)160

UE velocity (km/h)60

A3-Event Handover Procedure

Complexity:

O(N_{\text{neighbours}})

per measurement period, where

N_{\text{neighbours}}

is the number of monitored cells (typically 8--32 in LTE/NR).

Input: RSRP measurements,

H_{\mathrm{hys}}

,

T_{\mathrm{TTT}}

, filter coefficient

k

Initialise: timer

\leftarrow

inactive, serving cell

\leftarrow s

1. for each measurement period do

2.

\quad

Measure RSRP from serving cell

s

and neighbours

\{n_j\}

3.

\quad

Apply L3 filtering:

\hat{P}_j \leftarrow (1-2^{-k/4})\hat{P}_j + 2^{-k/4} P_j^{\mathrm{raw}}

4.

\quad

for each neighbour

n_j

do

5.

\quad\quad

if

\hat{P}_{n_j} > \hat{P}_s + H_{\mathrm{hys}}

then

6.

\quad\quad\quad

if timer inactive then start timer for

n_j

7.

\quad\quad\quad

if timer for

n_j

expired (

\geq T_{\mathrm{TTT}}

) then

8.

\quad\quad\quad\quad

Send Measurement Report to network

9.

\quad\quad\quad\quad

Network initiates handover to

n_j

10.

\quad\quad\quad\quad

Execute handover;

s \leftarrow n_j

11.

\quad\quad\quad

end if

12.

\quad\quad

else reset timer for

n_j

13.

\quad\quad

end if

14.

\quad

end for

15. end for

In 5G NR, conditional handover (CHO) prepares multiple target cells in advance, reducing the handover execution time and failure probability at high velocities.

Example: Handover Parameter Optimisation

A UE travels at $v = 60$ km/h through a network with cell radius $R = 500$ m and shadow fading standard deviation $\sigma_{\text{sf}} = 8$ dB.

(a) Compute the expected handover rate. (b) With $H_{\text{hys}} = 3$ dB, estimate the ping-pong probability (using $p_{\text{pp}} \approx Q(H_{\text{hys}}/\sigma_{\text{sf}})$ as a first-order approximation). (c) If $T_{\text{TTT}} = 160$ ms, compute the distance travelled during the TTT and assess whether this is acceptable for the given cell size. (d) Recommend parameters for a high-speed scenario ( $v = 120$ km/h).

Solution

Handover rate

(a) $v = 60$ km/h $= 16.67$ m/s.

$\mu_{\text{HO}} = \frac{2 \times 16.67}{\pi \times 500} = \frac{33.33}{1570.8} = 0.0212$ HO/s

About 1 handover every 47 seconds, or 76 HO/hour.

Ping-pong probability

(b) $p_{\text{pp}} \approx Q(3/8) = Q(0.375) \approx 0.354$ .

About 35% of handovers would be ping-pong with 3 dB hysteresis — this is unacceptably high.

TTT distance

(c) Distance during TTT: $d = v \cdot T_{\text{TTT}} = 16.67 \times 0.16 = 2.67$ m.

This is only 0.5% of the cell radius — very small, leaving ample margin for handover completion.

High-speed recommendations

(d) At 120 km/h ( $v = 33.33$ m/s):

HO rate doubles to $\sim$ 0.042/s (152 HO/hour)
Distance during TTT: 5.33 m (still $<$ 1.1% of $R$ )
Recommend increasing $H_{\text{hys}}$ to 5 dB to reduce $p_{\text{pp}}$ : $Q(5/8) = Q(0.625) \approx 0.266$
Consider reducing $T_{\text{TTT}}$ to 100 ms to prevent late handovers at high speed. $\blacksquare$

Quick Check

What is the effect of increasing the time-to-trigger $T_{\text{TTT}}$ in the A3 handover event?

It reduces both the ping-pong rate and the handover failure rate

It reduces the ping-pong rate but increases the risk of handover failure

It has no effect on handover performance

It reduces the handover rate proportionally

Correction:

It reduces the ping-pong rate but increases the risk of handover failure

A longer $T_{\text{TTT}}$ requires the A3 condition to hold for more time before triggering, filtering out transient fluctuations (reducing ping-pong). However, this delays the handover, and at high velocities the UE may leave the source cell's coverage before the handover completes.

Common Mistake: Handover Parameters Must Be Adapted to Velocity

Mistake:

Using fixed $H_{\text{hys}} = 3$ dB and $T_{\text{TTT}} = 160$ ms for all users regardless of their mobility state.

Correction:

At high speed ( $> 60$ km/h), the time the UE spends in the handover overlap zone is shorter, requiring faster handover execution. 3GPP NR defines mobility states (normal, medium, high) with speed-dependent parameter scaling:

Normal ( $< 30$ km/h): $H_{\text{hys}} = 3$ dB, $T_{\text{TTT}} = 320$ ms
Medium ( $30$ -- $60$ km/h): $H_{\text{hys}} = 2$ dB, $T_{\text{TTT}} = 160$ ms
High ( $> 60$ km/h): $H_{\text{hys}} = 1$ dB, $T_{\text{TTT}} = 80$ ms

Additionally, 5G NR introduces conditional handover (CHO) that pre-prepares multiple target cells, reducing handover interruption time from 40--60 ms to 0 ms (make-before-break).

Handover (Handoff)

The process of transferring an active connection from one cell to another as the user moves. In LTE/NR, the A3 event triggers handover when the target cell's RSRP exceeds the source cell's RSRP by the hysteresis margin for the time-to-trigger duration.

Related: Ping-Pong Handover, A3 Event

Ping-Pong Handover

A handover followed by a rapid return handover to the original cell, typically caused by shadow fading fluctuations or insufficient hysteresis margin. Ping-pong wastes radio resources, increases signalling load, and degrades user experience through repeated interruptions.

Related: Handover (Handoff)

A3 Event

A 3GPP-defined measurement event where the neighbour cell's measured signal strength exceeds the serving cell's by a configured hysteresis margin. When the A3 condition persists for the time-to-trigger duration, the UE sends a measurement report and the network initiates handover.

Handover and Mobility