Sensing Assisted Mobility Management for ISAC

Introduction

Mobility management in cellular networks encompasses handover (HO) decision-making, cell (re)selection, conditional handover (CHO), and dual-active protocol stack (DAPS) procedures. In 5G NR, these procedures rely on UE measurement reports (RSRP, RSRQ, SINR) of the serving and neighboring cells, configured via measurement events (A1–A6, B1–B2). The UE periodically or event-triggered reports measurements, and the network decides when and to which cell a handover should occur.

Sensing-assisted mobility management augments this process by using radar-like sensing at the gNB to track UE positions, velocities, and trajectories in real time. This additional information enables the network to make proactive and predictive handover decisions rather than reactive ones.

 Conventional vs. Sensing-Assisted Handover Decision
┌───────────────────────────────────────────────────────────────────────┐
│                                                                       │
│  Conventional (Reactive):                                             │
│                                                                       │
│  UE moves ──▶ RSRP drops ──▶ Event A3 ──▶ Meas Report ──▶ HO Cmd    │
│               (too late?)     triggered     (delay!)       (late!)    │
│                                                                       │
│  Sensing-Assisted (Proactive):                                        │
│                                                                       │
│  Sensing ──▶ Track UE ──▶ Predict ──▶ Pre-configure ──▶ Seamless     │
│  echoes     position     trajectory   target cell       HO           │
│             & velocity               (early!)                         │
│                                                                       │
└───────────────────────────────────────────────────────────────────────┘

Key Concepts

UE Trajectory Prediction

Sensing provides continuous position and velocity estimates of UEs (or UE-associated objects like vehicles). Using Kalman filtering or similar prediction algorithms, the gNB can:

Predict when the UE will cross the cell boundary.
Estimate which neighboring cell the UE is heading towards.
Determine the optimal handover preparation time.

 Trajectory-Based Handover Prediction
┌───────────────────────────────────────────────────────────────────┐
│                                                                   │
│          Cell A                    │            Cell B             │
│                                   │                               │
│      gNB_A ●                      │  (cell boundary)    ● gNB_B  │
│                                   │                               │
│           UE trajectory:          │                               │
│        ●───●───●───●───●──────▶───┼────▶                         │
│        t₀  t₁  t₂  t₃  t₄        │  t₅ (predicted crossing)     │
│                                   │                               │
│     Sensing tracks positions      │  HO prep starts at t₃        │
│     at t₀..t₄                     │  (2 steps ahead!)            │
│                                   │                               │
└───────────────────────────────────────────────────────────────────┘

Handover Failure Reduction

Handover failures occur due to:

Too-late handover: UE moves out of serving cell coverage before HO completes.
Too-early handover: UE is handed over to a cell it quickly leaves.
Ping-pong: UE oscillates between two cells.

Sensing-assisted mobility management addresses these by:

Velocity-aware hysteresis: Adjusting measurement event thresholds based on UE speed.
Direction-aware cell selection: Choosing the target cell aligned with the UE’s direction of travel.
Predictive CHO configuration: Pre-configuring conditional handover with sensing-predicted candidate cells.

Mobility State Estimation Enhancement

In LTE/NR, the UE performs mobility state estimation (MSE) by counting handover events. Sensing provides a direct and more accurate velocity estimate:

Table 8 Mobility State Detection Comparison
Metric	Conventional MSE	Sensing-Assisted MSE
Input	HO event count	Doppler-based velocity
Accuracy	Coarse (3 states)	Fine-grained (continuous)
Latency	Multiple HO events needed	Instantaneous from echo
Reliability	Cell-size dependent	Independent of cell layout

Connected-Mode Mobility Optimization

Beyond handover decisions, sensing assists other mobility procedures:

L1/L2 inter-cell mobility: Sensing can accelerate lower-layer cell switching by pre-identifying the target cell’s beam, reducing the L1/L2 handover interruption time.
Beam failure recovery: Sensing detects blockage events (e.g., a vehicle passing between the gNB and UE) and triggers beam failure recovery proactively.
Fast RLF detection: Sensing can predict radio link failure when a fast-moving UE is heading into a coverage hole identified in the sensing-derived environment map.

Benefits

Table 9 Benefits of Sensing-Assisted Mobility Management
Benefit	Description
Reduced HO failure rate	Predictive decisions prevent too-late and too-early handovers.
Lower HO interruption time	Early preparation of target cell resources shortens the handover gap.
Reduced ping-pong	Direction-aware cell selection avoids unnecessary handovers.
Improved high-speed UE support	Velocity-aware parameter tuning adapts to high-mobility scenarios.

Challenges

UE identification: Sensing detects targets but may not directly identify which target corresponds to which UE. Data association between sensing targets and RRC-connected UEs is needed.
NLOS mobility: In urban environments, UEs may not be directly visible to the gNB’s sensing subsystem due to blockage, limiting trajectory prediction accuracy.
Privacy concerns: Tracking UE positions via sensing raises user privacy considerations that may require anonymization or consent mechanisms.
Multi-cell coordination: Sensing-assisted handover may require sensing information exchange between neighboring gNBs, adding Xn interface signaling overhead.