top of page
Search

Link Adaptation in 5G NR

  • Writer: Venkateshu
    Venkateshu
  • 6 days ago
  • 5 min read

1. Introduction

Link adaptation is a core feature in 5G NR that enables dynamic adjustment of transmission parameters — mainly the Modulation and Coding Scheme (MCS) — based on real-time wireless channel conditions. This process ensures optimal throughput and reliability for each User Equipment (UE) by adapting to mobility, interference, fading, and changing network scenarios.

Modern 5G networks advance beyond 4G LTE by employing smarter, faster, and more flexible link adaptation algorithms fueled by machine learning and data-driven decision-making.

 

2. Fundamentals of Link Adaptation

Key Goals:

  • Maximize spectral efficiency (bits/Hz)

  • Deliver high reliability (target block error rate, usually about 10%)

  • Respond to rapid changes in SINR and radio environment

Core Link Adaptation Process:

  1. Channel Quality Measurement: UE measures link quality using reference signals (CSI-RS in 5G NR).

  2. Reporting: UE sends Channel State Information (CSI) to gNB, including CQI (Channel Quality Indicator), RI (Rank Indicator), PMI (Precoding Matrix Indicator).

  3. MCS Selection: gNB maps CQI to suitable MCS, balancing throughput and reliability. MCS governs modulation order (QPSK, 16QAM, 64QAM, 256QAM) and coding rate.

  4. Feedback Loops: HARQ (Hybrid ARQ) feedback (ACK/NACK) informs continuous adjustments in MCS.

Example:

  • If a UE at cell edge experiences fading and reports low CQI, gNB shifts the transmission to a more robust MCS (lower modulation, stronger coding).

  • If a UE moves closer and CQI increases, gNB selects higher-order MCS for higher throughput.

 Adaptive Modulation and Coding (AMC)

Link adaptation is a higher-level control process that includes AMC as a part. It uses all available feedback—CQI from UE, as well as HARQ ACK/NACKs—to continuously refine ("adapt") the choice of modulation, coding, and (if supported) sometimes power or transmission rank.

 

  • Definition: AMC is the PHY-layer technique where the system dynamically selects the modulation scheme (e.g., QPSK, 16QAM, 64QAM, 256QAM) and coding rate based on the instantaneous channel conditions. It aims to maximize throughput while keeping error rates within acceptable limits.

  • How It Works: For a given SNR (provided by the UE as CQI), the appropriate Modulation and Coding Scheme (MCS) is chosen so that the block error rate (BLER) remains under target (e.g., 10%).

  • Example: If a UE experiences a high SNR, the network might select 256QAM with a high coding rate. If the channel degrades, it might switch to 16QAM with a lower code rate to improve robustness.

  • Output: MCS index, which directly determines PHY-layer configuration for each transmission.


ree

 

 Link adaptation in 5G NR comprises:

  • Mapping CQI to MCS (the AMC step)

  • Outer loop control (OLLA), using HARQ feedback to adjust AMC mapping so BLER target is met under real-world, sometimes imperfect CQI/SNR estimates

  • Adjusting link parameters such as power offsets, rank, beam selection, etc., when needed

  • Features: Incorporates feedback loops (inner loop = fast AMC mapping, outer loop = slower HARQ-based tweaks), can utilize machine learning for smarter decisions.

  • Output: The actual, continually updated PHY-layer parameters (including MCS from AMC), providing robust error control over time.

Aspect

Adaptive Modulation & Coding (AMC)

Link Adaptation

Scope

Selection of modulation and coding for a given instant

End-to-end process of adjusting link parameters

Feedback

Primarily uses SNR/CQI

CQI + HARQ ACK/NACK + historical data

Loop

"Inner loop" (immediate)

Both inner and outer loops

Role

Converts link quality to optimal MCS index

Ensures MCS mapping delivers BLER target, adapts over time

Level

PHY, per transmission

Cross-layer (PHY + MAC), continuous

 

3. Types of Link Adaptation Mechanisms

(a) Inner Loop Link Adaptation (ILLA)

ILLA handles fast, direct adjustment based on each UE's reported Channel Quality Indicator (CQI):

  • UE measures downlink quality (e.g., using CSI-RS).

  • UE reports CQI to gNB.

  • gNB maps CQI (via static look-up table shown below) to an MCS index for the next transmission.

  • This mapping reflects the link condition estimate for that time slot/TTI.

ree

Example:

  • UE measures CSI-RS and reports CQI=10.

  • gNB maps CQI=10 to MCS=17.

  • Transport block for next slot is assembled accordingly.

Pro: Fast adaptation to channel changes.Con: Vulnerable to errors/noise in CQI; can drift from BLER target if channel is non-ideal or feedback is imperfect.

 

(b) Outer Loop Link Adaptation (OLLA)

OLLA provides a feedback mechanism to fine-tune the MCS target, compensating for real link performance observed through HARQ ACK/NACK responses:

  • After transmission, the gNB receives ACK (success) or NACK (failure).

  • If BLER is above the configured target (e.g., 10%), OLLA adjusts a correction offset (Δoffset) downward (less aggressive MCS). If BLER is below target, offset nudges upward (more aggressive MCS).

  • This offset is added to the SINR→CQI mapping in ILLA, ensuring over time BLER converges to target—even with imperfect input.

Pro: Maintains robust, stable BLER and adapts to slowly changing systematic errors in SINR/CQI reporting.

Con: Response is slower—optimal settings for step size (Δup, Δdown) are a tradeoff between stability and responsiveness.


ree

 

 This illustrates:

  • ILLA rapidly reacts to current CQI value (plus OLLA offset).

  • OLLA slowly notches the offset to steer BLER toward the configured target.

Detailed Example: Offset Adjustment in OLLA

Let’s say the BLER target is 10%. The OLLA works as follows:

  • For each received HARQ NACK, decrease offset by Δdown (e.g., -0.1 dB).

  • For each ACK, increase offset by Δup (e.g., +0.03 dB).

  • The ratio Δup/Δdown ≈ BLERtarget/(1−BLERtarget) to make offset drift toward a stable target.

Numerical Example:

  • Initially: OLLA offset = 0. gNB maps CQI=12, MCS=22.

  • If UEs NACK frequently, offset decreases. This means next time CQI=12 is reported, the gNB will select a lower, more robust MCS (say, MCS=21).

  • As more ACKs come in, offset increases and MCS gets more aggressive.

Real-Time Example:

  • UE1: CQI=15, OLLA offset +0.2 dB (recent good channel), scheduled with MCS=28 on 10 PRBs.

  • UE2: CQI=7, OLLA offset −0.35 dB (fading), scheduled with MCS=10 on 20 PRBs.

  • Both get stable BLER ≈ 10% while maximizing cell throughput.

 

How Link Adaptation Supports MAC Layer Scheduling

The MAC scheduler in gNB leverages link adaptation results (current CQI/MCS) to:

  • Decide which users get which resource blocks (PRBs) and at what transport block sizes.

  • Balance throughput and fairness — e.g., cell-edge users may get more PRBs with robust (low MCS) assignment; near-cell users get fewer PRBs at higher MCS for peak rates.

  • Optimize QoS by prioritizing users based on latency, BLER targets, and buffer size.

  • Adjust scheduling rapidly as the link adaptation algorithm updates each user’s achievable rate after every CSI report.

Example Scenario:

  • UE1 (near gNB): Reports CQI = 15 → gNB chooses MCS = 28 (256QAM, high coding rate). UE1 scheduled on fewer PRBs, but achieves max throughput.

  • UE2 (cell edge): Reports CQI = 5 → gNB selects MCS = 6 (QPSK, low coding rate). To maintain fairness/QoS, MAC scheduler allocates more PRBs/time to UE2.

 

4. Algorithms: LTE vs. 5G NR Link Adaptation

Feature

LTE

5G NR

CSI feedback

Limited (mainly CQI); periodic

Rich reports: CQI, PMI, RI, CRI; slot-level freq.

MCS mapping

Fixed tables; OLLA

Flexible tables; OLLA + RL/AI; vendor-optimized

Adaptation speed

TTI-level (1 ms); moderate reaction

Slot-level (down to 0.125 ms); rapid adjustment

Scheduler integration

Basic input

Deep integration (MAC logic, ML forecasts)

Mobility adaptation

Rigid (hard to forecast)

ML/AI enables prediction of mobility patterns

Transport block size

Direct from MCS, RB

Dynamic (MCS, RB, numerology, slot size, layers)

Key Advances in 5G NR:

  • Uses enhanced CSI feedback and frequent slot-based adaptation.

  • RL/ML algorithms (like DDPG) enable smarter, context-aware decisions that boost throughput and reduce BLER.

  • Integration of link adaptation into MAC scheduling logic for smart resource allocation.

 

5. Summary Table: Link Adaptation Evolution

Aspect

LTE

5G NR

Feedback Loop

ILLA, OLLA

ILLA, OLLA, RL/ML-informed

Adaptation Speed

TTI (1 ms)

Slot (0.125–1 ms); faster, more flexible

CQI Mapping

Fixed reference tables

Vendor-optimized; dynamic, RL-based flex mapping

MAC Scheduler Use

Basic: MCS input to scheduling

Integrated: link adaptation logic drives resource allocation

 

6. Conclusion

Link adaptation is pivotal for high-performance, reliable 5G NR communications. Its evolution from fixed-table, slow-feedback LTE models toward flexible, RL/ML-enabled frameworks in 5G NR allows modern schedulers to respond instantly, optimize resources intelligently, and meet the demands of enhanced mobile broadband (eMBB), low-latency (URLLC), and massive machine communication (mMTC).

Through real-time CQI reports, advanced learning algorithms, and tight MAC integration, link adaptation ensures that every UE gets optimal throughput and reliability — whether stationary, moving, or at the cell edge.

 

7. References

 

 

​

 

bottom of page