SR state machine, SR failure → RA fallback, 2-step RACH & 6G predictive grants
Pre-allocated periodic resources. Zero latency for grant acquisition. Best for URLLC/VoNR.
1-bit signal on PUCCH. "I need a grant!" — 1–4 ms latency. Best for bursty traffic.
4-step or 2-step RACH. Always works. 5–20 ms latency. Fallback when SR fails.
SR is a 1-bit signal on PUCCH (Format 0 or Format 1). It tells the gNB "I have data" without saying how much — that's the BSR's job, sent later in the grant.
| Parameter | PUCCH Format 0 | PUCCH Format 1 |
|---|---|---|
| Duration | 1–2 symbols (short) | 4–14 symbols (long) |
| Capacity | 1–2 bits | 1–2 bits |
| Latency | Lower | Higher |
| Coverage | Limited | Better (longer) |
| Use case | Cell center, low latency | Cell edge, coverage limited |
Watch the SR state machine in action. Trigger data arrival, see SR transmissions with counter, and observe success or failure → RA fallback.
| State Variable | Initial Value | Purpose |
|---|---|---|
| SR_COUNTER | 0 | Number of SR transmissions so far |
| SR_PENDING | false | Whether SR is currently triggered |
| sr-ProhibitTimer | stopped | Min time between consecutive SRs |
| sr-TransMax | RRC configured | Max SR attempts before RA fallback |
When SR_COUNTER > sr-TransMax, MAC notifies RRC to release PUCCH/SRS resources, then initiates contention-based Random Access. Msg3 carries the BSR — the same information that would have been sent if SR had succeeded.
Animated timeline showing SR attempts, failure detection, and RA fallback (4-step). Adjust sr-TransMax to see how many retries before fallback.
PRACH uses broadcast preamble pool — always available. No PUCCH config needed. Works even after radio link failure.
RA is slower (10–20 ms vs 1–4 ms for SR), uses more resources (4 messages), and has collision risk. But it ALWAYS works.
Releases PUCCH SR resources and SRS resources. UE can't transmit on UL until RA completes.
| Aspect | 4-Step RACH | 2-Step RACH |
|---|---|---|
| Messages | Msg1 + Msg2 + Msg3 + Msg4 | MsgA + MsgB (combined) |
| MsgA/Msg1 | PRACH preamble only | PRACH preamble + PUSCH payload (BSR + ID) |
| Latency | 12–20 ms | 5–10 ms |
| Collision handling | Msg4 resolves | MsgB resolves (fallback to 4-step if fail) |
| Signal requirement | Any RSRP | RSRP > threshold (strong signal) |
Side-by-side animated message flow. Watch the latency difference — 2-step completes in roughly half the time.
| Periodicity | Avg Wait | SR→Grant | Total Latency | Use Case |
|---|---|---|---|---|
| sl1 (1 slot) | 0.5 ms | ~1 ms | ~1.5 ms | URLLC |
| sl2 (2 slots) | 1 ms | ~1 ms | ~2 ms | Voice (VoNR) |
| sl5 (5 slots) | 2.5 ms | ~1 ms | ~3.5 ms | Video |
| sl20 (20 slots) | 10 ms | ~1 ms | ~11 ms | Web |
| sl80 (80 slots) | 40 ms | ~1 ms | ~41 ms | Background |
Adjust the SR periodicity slider to see how average wait time changes. Shorter periodicity = lower latency but more PUCCH overhead.
Bursty, unpredictable traffic. 2–5 ms latency. PDCCH overhead per grant. gNB decides per-TTI. Best for eMBB, web, video.
Periodic, predictable traffic. 0.5–2 ms latency. No PDCCH overhead. Fixed pre-allocated resources. Best for VoNR, URLLC, V2X.
CG for periodic VoNR packets (every 20 ms). SR for bursty web/video data. RA fallback if SR fails. Best of both worlds.
| Scenario | sr-Periodicity | sr-TransMax | sr-Prohibit |
|---|---|---|---|
| VoNR (voice) | sl2 (2 slots) | n8 | ms2 |
| Video streaming | sl5 (5 slots) | n16 | ms5 |
| Web browsing | sl20 (20 slots) | n32 | ms10 |
| IoT background | sl80 (80 slots) | n64 | ms40 |
| URLLC | sl1 (1 slot) | n4 | ms1 |
AI at gNB predicts when UE will have data → pre-sends grant before data arrives → zero SR delay. Targets 60–80% SR reduction.
1-symbol "demand signal" → gNB responds within same slot. No pre-configured SR resources. Total latency < 1 slot.
Single RA procedure handles all access types. AI selects 2-step vs 4-step, CBRA vs CFRA. Adaptive to channel, urgency, load.
Compare SR/grant-acquisition latency from 4G through 6G with animated timeline.
| Dimension | 4G LTE | 5G NR | 6G (Target) |
|---|---|---|---|
| SR format | PUCCH F1 only | Format 0 + Format 1 | + Demand signal |
| Min SR latency | 1 ms | 0.125 ms | < 0.1 ms |
| SR configs per UE | 1 | 8 | AI-adaptive |
| RA fallback | 4-step only | 4-step + 2-step | Unified RA |
| Predictive grant | None | None | AI predicts |