Link Adaptation & HARQ

Link Adaptation dynamically adjusts transmission parameters based on real-time channel conditions. Combined with HARQ and its round-trip timing, it forms the closed-loop system that maximizes spectral efficiency in 5G NR. Covers MCS selection, CQI reporting, OLLA, and HARQ RTT.

Channel Coding Page → Spatial Mux Page →

#1The Link Adaptation Loop

The link adaptation loop tracks the channel and adjusts transmission parameters every slot (or few slots):

Channel Speed: Normal BLER Target: 10%

Real-time link adaptation — watch MCS, CQI, and throughput track channel conditions slot-by-slot

What Gets Adapted?

Parameter	What Changes	Controlled By	Speed
Modulation order	QPSK → 16QAM → 64QAM → 256QAM	MCS index	Per-slot
Code rate	0.076 → 0.926	MCS index	Per-slot
Layers	1 → 4 spatial streams	RI report	Per-slot
PRBs	1 → 273 PRBs	Scheduler	Per-slot
Beam direction	SSB beam selection	Beam mgmt	Per-slot

#2MCS — Modulation and Coding Scheme

The MCS index simultaneously specifies modulation order and target code rate. It is the primary output of link adaptation:

MCS Index: 14 Layers: 2

MCS table — drag to see modulation, code rate, and spectral efficiency change in real-time

MCS Tables — 3 Options

Table	Max Modulation	Used For	MCS Range
Table 1 (Default)	256QAM	eMBB data	0–27
Table 2 (Low SE)	64QAM	URLLC	0–27
Table 3 (Coverage)	64QAM	Coverage enhancement	0–28

#3CQI — Channel Quality Indicator

The UE reports a CQI index (1–15) that maps to the highest modulation/code rate it can decode at ≤ 10% BLER:

SINR: 10.0 dB

CQI selection — UE finds the highest CQI where estimated BLER ≤ 10% at the measured SINR

CQI Table (TS 38.214)

CQI	Modulation	Code Rate	SE (bps/Hz)	≈ SNR
1	QPSK	0.117	0.234	-6 dB
4	QPSK	0.439	0.877	0 dB
7	16QAM	0.379	1.516	5 dB
9	16QAM	0.602	2.406	9 dB
11	64QAM	0.554	3.322	13 dB
13	64QAM	0.754	4.523	17 dB
15	256QAM	0.926	7.406	22 dB

CQI Reporting Modes

Mode	Trigger	Channel	Granularity
Periodic (P-CSI)	RRC configured	PUCCH	Wideband only
Aperiodic (A-CSI)	DCI request	PUSCH	Subband OK
Semi-Persistent (SP-CSI)	MAC CE	PUCCH/PUSCH	Configurable

#4Transport Block Size (TBS)

  TBS Determination (TS 38.214 §5.1.3.2):

  Step 1: N_info = N_RE × R × Q_m × ν
  Step 2: If N_info ≤ 3824:
            TBS = max(24, 2^n × round(N_info / 2^n))
          If N_info > 3824:
            TBS = max(3840, 2^n × round((N_info-24) / 2^n)) + 24
  Step 3: If TBS > 8448 → segment into C code blocks

  Example: MCS=20, 50 PRBs, 2 layers
    N_RE = 50×132 = 6600
    N_info = 6600 × 0.7109 × 6 × 2 = 56,303
    TBS ≈ 55,320 bits → 7 code blocks

#5Outer Loop Link Adaptation (OLLA)

OLLA adjusts MCS selection to track the actual BLER toward the target (10%):

On ACK (success)

OLLA_offset += Δ_up (small step, e.g., 0.01 dB)
Increase MCS — be more aggressive

On NACK (failure)

OLLA_offset -= Δ_down (large step, e.g., 0.1 dB)
Decrease MCS — be more conservative

Step Size Ratio

Δ_down / Δ_up = (1 - BLER_target) / BLER_target
For 10% target: Δ_down/Δ_up = 0.9/0.1 = 9 → convergence guaranteed

Channel Noise: 30%

OLLA convergence — watch the offset adapt slot-by-slot to maintain target BLER ≈ 10%

#6HARQ Round-Trip Time (RTT)

The HARQ RTT is the time between initial transmission and earliest possible retransmission. It determines pipeline depth and throughput efficiency:

K1 (PDSCH→ACK): 3 slots K_sched (delay): 2 slots Numerology μ: 1 (30kHz)

HARQ timeline — adjust K1 and scheduling delay to see RTT impact on pipeline utilization

Key Timing Parameters

Parameter	Definition	Typical Range
K1	PDSCH → HARQ-ACK delay (slots)	1–15 (typically 2–4)
N1	UE PDSCH processing time (symbols)	3–20 (depends on UE cap)
K2	DCI → PUSCH delay (slots)	1–36 (typically 2–4)
N2	UE PUSCH preparation time (symbols)	5–36 (depends on UE cap)

Pipeline Utilization

Utilization = min(1, N_HARQ / HARQ_RTT)
If RTT = 5 slots, 16 processes → 31% utilization. If RTT = 16 slots → 100% (full pipeline). RTT > 16 slots → pipeline stall → throughput drops.

#7HARQ-ACK Multiplexing & Bundling

Multiplexing (per-process)

Each HARQ process gets its own ACK/NACK bit. Payload: [ACK, NACK, ACK] = [1,0,1]. Precise but higher overhead.

Bundling (AND logic)

Multiple ACKs are AND-combined. If any failed → NACK for all. Lower overhead, more conservative.

PUCCH Format	Max Bits	Used For
0	1–2	1–2 HARQ-ACK (no SR/CSI)
1	1–2	1–2 bits with SR
2	> 2	Multiple ACK + CSI + SR
3	> 2	Large payload (long PUCCH)

#8New Data Indicator (NDI)

The NDI is a 1-bit DCI field that toggles to distinguish new data from retransmissions:

NDI Toggle Rule

NDI_current ≠ NDI_previous → NEW transmission (flush soft buffer)
NDI_current = NDI_previous → RETRANSMISSION (combine with stored LLRs)

  Example HARQ process #3:
  TX #1: NDI=0 → UE: "toggled → new data" → flush buffer, decode
  TX #2: NDI=0 → UE: "no toggle → reTX"    → combine + decode
  TX #3: NDI=1 → UE: "toggled → new data"  → flush buffer, decode

#9Link Adaptation Performance Metrics

KPI	Target	Too High	Too Low
First-TX BLER	10%	>20% → MCS too aggressive	<3% → MCS too conservative
Residual BLER	<0.1%	>1% → HARQ not recovering	—
ReTX Rate	<15%	>25% → excessive overhead	<3% → wasting capacity
CQI Accuracy	±1 index	Monitor vs optimal CQI (offline analysis)

#10Practical Field Considerations

Pitfall	Impact	Best Practice
OLLA not converged	Wrong MCS → reTX storm or wasted capacity	Monitor OLLA offset; ±2 dB steady-state
Stale CQI in mobility	MCS mismatch → BLER spike	Short CQI period (<20 ms) for mobile UEs
K1 too small	ACK timing violation	Verify K1 ≥ N1 for UE capability
HARQ pipeline stall	Throughput drops 20-40%	Reduce K1 or add HARQ processes
Ignoring subband CQI	Missing 10-30% freq-selective gain	Enable aperiodic subband CQI for cell-center
Wrong BLER target	Over/under-provisioning	Standard: 10% first-TX BLER

LTE vs 5G NR Link Adaptation

Aspect	LTE	5G NR
MCS tables	1 table	3 tables (64QAM/256QAM/low-SE)
HARQ processes	8 (FDD)	16
HARQ timing	Fixed (4 subframes FDD)	Flexible (K1/K2 per-slot)
Min HARQ RTT	8 ms	2-5 slots (0.5-2.5 ms)
CQI reporting	Periodic + aperiodic	+ semi-persistent

#11Further Reading

3GPP Specifications

TS 38.214 — NR Physical Layer Procedures for Data (MCS, TBS, CQI, HARQ)
TS 38.213 — NR Physical Layer Procedures (HARQ-ACK timing, K1, K2)
TS 38.212 — NR Multiplexing and Channel Coding

Companion Resources

Channel Coding Page → Spatial Multiplexing Page → ☆ AI Scheduling → ☆ NTN MAC →