Sensing Assisted Beamforming for ISAC
Introduction
In conventional communication systems, beamforming relies on pilot-based channel estimation and codebook-based feedback to steer transmit and receive beams towards the intended user. This process involves significant overhead due to beam sweeping, CSI-RS transmission, and feedback reporting, especially for millimeter-wave (mmWave) and sub-THz bands where narrow beams are required and the beam search space is vast.
Sensing-assisted beamforming leverages radar-like sensing information — such as angle of arrival (AoA), angle of departure (AoD), range, and velocity of targets — to predict or refine beam directions without exhaustive beam sweeping. By exploiting the spatial information extracted from sensing echoes, the base station (gNB) can narrow down the beam search space, reduce beam alignment latency, and improve beam tracking for mobile users.
Sensing-Assisted Beamforming: High-Level Flow
┌─────────────────────────────────────────────────────────────────────┐
│ │
│ ┌──────────────┐ ┌──────────────────┐ ┌──────────────────┐ │
│ │ Sensing │ │ Spatial Info │ │ Beam Candidate │ │
│ │ Signal │───▶│ Extraction │───▶│ Narrowing │ │
│ │ (Echo) │ │ (AoA/AoD/Range) │ │ (Reduced Set) │ │
│ └──────────────┘ └──────────────────┘ └───────┬──────────┘ │
│ │ │
│ ▼ │
│ ┌──────────────┐ ┌──────────────────┐ ┌──────────────────┐ │
│ │ Beamformed │ │ Beam Refinement │ │ Beam Selection │ │
│ │ Data TX/RX │◀───│ (Fine Alignment)│◀───│ (Best Beam) │ │
│ └──────────────┘ └──────────────────┘ └──────────────────┘ │
│ │
└─────────────────────────────────────────────────────────────────────┘
Key Concepts
Beam Prediction from Sensing Parameters
Sensing echoes provide estimates of the target’s angular position (AoA/AoD) and range. These parameters can be directly mapped to a subset of candidate beams, bypassing the need for full beam sweeping. The mapping is:
AoA/AoD → Beam index: The angular estimate from sensing narrows the beam search to a small angular window.
Range → Timing advance: The range estimate helps predict the appropriate timing advance for the UE.
Doppler → Beam tracking rate: Velocity information helps decide how frequently beam re-alignment is needed.
Beam Tracking using Sensing
For mobile UEs, beam tracking is critical to maintain alignment as the UE moves. Conventional beam tracking uses periodic CSI-RS measurements and beam reports, which consume significant overhead. Sensing-assisted beam tracking exploits the continuous availability of radar echoes to predict the UE’s trajectory and proactively update the serving beam.
Sensing-Assisted Beam Tracking
┌──────────────────────────────────────────────────────────────────┐
│ │
│ Time t₀ Time t₁ Time t₂ Time t₃ │
│ ┌──────┐ ┌──────┐ ┌──────┐ ┌──────┐ │
│ │Sense │ │Sense │ │Sense │ │Sense │ │
│ │Echo │────────▶│Echo │────────▶│Echo │────────▶│Echo │ │
│ └──┬───┘ └──┬───┘ └──┬───┘ └──┬───┘ │
│ │ │ │ │ │
│ ▼ ▼ ▼ ▼ │
│ AoA=30° AoA=32° AoA=34° AoA=36° │
│ Beam #5 Beam #5 Beam #6 Beam #6 │
│ (switch!) │
│ │
│ ──────────── Trajectory Prediction ──────────────▶ │
│ │
└──────────────────────────────────────────────────────────────────┘
Hierarchical Beam Search Reduction
In massive MIMO systems with large antenna arrays, the beam codebook can contain hundreds of beams. A hierarchical beam search first uses wide beams, then narrows down. Sensing can assist by:
Skipping the wide-beam stage entirely if sensing AoA resolution is sufficient.
Starting from an intermediate codebook level based on the angular uncertainty of the sensing estimate.
Eliminating angular sectors where no sensing targets are detected.
Benefits
Benefit |
Description |
|---|---|
Reduced beam alignment latency |
Sensing narrows the search space from N beams to a small subset, cutting initial access time. |
Lower CSI-RS overhead |
Fewer CSI-RS resources needed when sensing pre-selects candidate beams. |
Improved beam tracking |
Continuous sensing echoes enable proactive beam updates without waiting for UE reports. |
Robustness to blockage |
Sensing can detect blockage events (sudden drop in echo power) faster than link-level monitoring. |
Multi-user beam management |
Sensing can simultaneously track multiple targets, aiding multi-user MIMO beam scheduling. |
Challenges
Sensing-communication beam mismatch: The sensing beam pattern and communication beam pattern may not be identical, especially if separate antenna panels or subarrays are used.
Angular resolution limits: Sensing angular resolution depends on the array aperture and bandwidth. For small arrays or narrow bandwidth, the angular estimate may not be precise enough to select a single beam.
NLOS environments: In rich scattering environments, sensing echoes may arrive from reflected paths that do not correspond to the direct path used for communication.
Latency of sensing processing: The sensing signal processing pipeline (matched filtering, angle estimation, CFAR detection) must be fast enough to provide beam updates within the coherence time.
3GPP Relevance
In 3GPP Rel-19 ISAC study item (SI), sensing-assisted beam management is one of the identified use cases under the category of “sensing-assisted communication.” The study considers how sensing measurements at the gNB can be used to reduce beam management overhead, particularly for FR2 (mmWave) deployments where beam alignment is most challenging. The key question under study is whether existing beam management procedures (P1/P2/P3) can be enhanced with sensing side-information or whether new procedures are needed.