N1 processing time, overlapping PDSCH, DL preemption (DCI 2_1) & 6G evolution
When gNB schedules downlink data, the UE receives a DL assignment via DCI (Format 1_0/1_1/1_2) on PDCCH, then receives PDSCH data, decodes it, and sends HARQ-ACK/NACK. The critical timing parameters are K0 (DCI-to-PDSCH offset) and N1 (PDSCH processing time before HARQ-ACK).
Slot offset between DCI and PDSCH. Configured by gNB via time-domain resource assignment table. Typically 0-3 slots.
Minimum symbols between end of PDSCH and start of HARQ-ACK. Depends on UE capability (Cap 1 vs Cap 2) and SCS.
Slot offset configured by gNB. Must satisfy K1 ≥ N1. Controls HARQ-ACK timing on PUCCH.
| SCS | N1 Cap 1 (symbols) | N1 Cap 2 (symbols) | Time (Cap 1) | Time (Cap 2) |
|---|---|---|---|---|
| 15 kHz | 8 | 3 | 571 μs | 214 μs |
| 30 kHz | 10 | 5 | 357 μs | 179 μs |
| 60 kHz | 17 | 9 | 304 μs | 161 μs |
| 120 kHz | 20 | 10 | 179 μs | 89 μs |
Visualize the DL assignment timeline: DCI → K0 → PDSCH → N1 processing → HARQ-ACK. Compare Cap 1 (slow) vs Cap 2 (fast). Adjust K1 to see valid/invalid configurations.
Multiple DL assignments can target the same time/frequency resources. The UE resolves overlaps using priority-based rules: higher PHY-PriorityIndex wins, then Dynamic > SPS, then earlier DCI wins.
Two PDSCH grants overlap in time+frequency. Watch the UE resolve which one to receive based on priority and grant type rules.
PHY-PriorityIndex 1 (URLLC) beats PriorityIndex 0 (eMBB). Higher-priority PDSCH is received, lower is dropped.
Same priority: Dynamic PDSCH beats SPS PDSCH. The dynamic grant overrides the pre-configured SPS occasion.
Only overlapping RBs/symbols are affected. Non-overlapping parts of both PDSCH are still received and decoded.
When urgent URLLC data arrives, gNB can puncture an ongoing eMBB PDSCH by transmitting URLLC on some of the same RBs. The UE is notified via DCI Format 2_1 (Preemption Indication), which tells it exactly which resources were preempted so it can zero-out those LLRs (soft erasure).
Watch an eMBB PDSCH get punctured by URLLC traffic. The DCI 2_1 bitmap highlights which RBs/symbols were preempted. Toggle preemption to see the impact on eMBB decoding.
| Metric | Without Preemption | With Preemption |
|---|---|---|
| eMBB BLER | ~10% | ~15-25% |
| eMBB throughput | Baseline | -5-15% |
| URLLC latency | 5-10 ms (wait) | < 1 ms |
| eMBB reTX rate | Normal | +10-20% |
Rel-16 introduced per-PDSCH priority levels: PriorityIndex 0 (normal/eMBB) and PriorityIndex 1 (high/URLLC). When overlapping PDSCH grants conflict, the UE deterministically resolves them by priority. HARQ-ACK for each priority uses separate PUCCH resource sets.
| Scenario | N1 Cap | K1 Setting | Preemption | Priority |
|---|---|---|---|---|
| eMBB (normal) | Cap 1 | K1 ≥ 2 | Monitor | Index 0 |
| URLLC (critical) | Cap 2 | K1 = 0-1 | Enable | Index 1 |
| VoNR DL | Cap 1 | K1 ≥ 2 | Not needed | Index 0 |
| IoT DL | Cap 1 | K1 ≥ 4 | Not needed | Index 0 |
| Aspect | LTE | 5G NR |
|---|---|---|
| Processing time | Fixed ~3 ms | N1 (variable, symbol-level) |
| DL preemption | Not supported | DCI Format 2_1 |
| PDSCH priority | Not supported | PHY-PriorityIndex (0/1) |
| Same-slot ACK | Not possible | Cap 2 (N1 = 3 sym) |
AI at UE learns scheduling patterns → skips monitoring in empty slots → 40-60% PDCCH monitoring reduction → major battery savings.
AI predicts URLLC arrivals → pre-reserves resources → no preemption needed → eMBB never punctured → DCI 2_1 eliminated.
Advanced silicon targets N1 < 50 μs (sub-symbol). Same-slot HARQ-ACK always possible. End-to-end DL latency < 0.5 ms.
Compare N1 processing time, preemption support, and priority features across generations.
| Dimension | 4G (LTE) | 5G NR | 6G (Target) |
|---|---|---|---|
| N1 processing | ~3 ms | 214 μs (Cap 2) | < 50 μs |
| DL preemption | None | DCI 2_1 | AI eliminates |
| PDSCH priority | None | Index 0/1 | Multi-level |
| PDCCH monitoring | Always | Configurable | AI-predicted |
| Same-slot ACK | No | Cap 2 only | Always |