Beam Tracking
Understanding Beam Tracking
Beam tracking is the steady-state mode of 5G NR beam management, where the system has already identified the optimal beam pair and must now maintain it as conditions change. The challenge is detecting when the current beam is no longer optimal and switching to a better one fast enough to avoid throughput degradation or beam failure.
The tracking loop balances measurement overhead against responsiveness. More frequent CSI-RS measurements provide faster detection of beam changes but consume more downlink resources. The gNB dynamically adjusts the measurement configuration based on UE mobility (estimated from Doppler spread or beam report variability), providing frequent measurements only when needed.
Beam Tracking Timeline
CSI-RS measurement: 1 slot (0.5 ms)
UE processing + reporting: 1–3 slots (0.5–1.5 ms)
gNB decision: 1 slot (0.5 ms)
MAC CE activation: 3 ms (after HARQ ACK)
Total: 4.5–5.5 ms (MAC CE path)
DCI path: 1.5–2.5 ms (same-slot)
Beam Dwell Time (FR2, 5° beam):
Coverage at 100 m: 2R·tan(2.5°) = 8.7 m
At 60 km/h: 8.7/16.7 = 520 ms
At 120 km/h: 8.7/33.3 = 260 ms
At 300 km/h: 8.7/83.3 = 104 ms
Min tracking rate = 2 × 1/(dwell time)
(Nyquist: measure at 2x beam change rate)
Tracking vs. Refinement vs. Recovery
| Feature | Beam Tracking | Beam Refinement | Beam Recovery |
|---|---|---|---|
| Mode | Connected (steady-state) | Post-access (one-time) | Post-failure |
| Candidates monitored | 1–4 beams | 4–8 beams | All candidates |
| Reference signal | Periodic CSI-RS / SSB | Aperiodic CSI-RS | SSB + CSI-RS |
| Update period | 5–500 ms | 10–40 ms (burst) | 20–50 ms (total) |
| Trigger | Periodic/event-based | Initial access / HO | BFD declaration |
Frequently Asked Questions
How does it work?
Closed loop: UE measures L1-RSRP on SSB/CSI-RS, reports best N beams, gNB compares to serving beam. If best exceeds current by 3 to 6 dB, TCI state update (MAC CE 3 ms or DCI same-slot). Total loop: 2 to 5.5 ms.
Speed requirements?
Driven by beam dwell time: 5° beam at 100 m = 8.7 m coverage. 60 km/h: 520 ms dwell, track every 200 ms. 300 km/h: 104 ms dwell, track every 50 ms. Blockage events faster: 50 to 200 ms.
Tracking vs. refinement?
Tracking: continuous steady-state, 1 to 4 candidates, periodic CSI-RS. Refinement: one-time narrowing, 4 to 8 candidates, aperiodic CSI-RS. Tracking detects when to switch; refinement optimizes after switching.