LDPC Decoder LLR Scaling Adaptation for Changing Channel Conditions

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Conventional LDPC decoder algorithms face challenges in achieving optimal LLR scaling due to channel and UE conditions, leading to wireless performance degradation, and existing solutions either fail to adapt to changing conditions or result in performance trade-offs.

Innovation Solution

A self-adapting LDPC decoder architecture that utilizes a machine-learning based LDPC decoder control input model to predict optimal scaling factors by learning from posteriori decoding metrics using unused LDPC decoder accelerators, adapting LLR scaling over time based on channel conditions.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Adaptability or versatility

If conventional LDPC decoder algorithms use fixed LLR scaling, then device complexity is reduced, but wireless performance degrades due to inability to adapt to changing channel conditions

Engineering Contradiction:
Improveadaptability to channel conditionsVSAvoiddecoder architecture complexity
Core Design Contradiction:
Adaptability or versatilityVSDevice complexity

Solution Approach 1:

The system uses idle LDPC decoder accelerators to automatically train and update the neural network model without external intervention. The unused decoders perform background training using historical LLR data, enabling the system to self-improve its scaling accuracy over time while maintaining operational simplicity.

Inventive Principle:
Principle #25Self-service

Solution Approach 2:

The patent dynamically changes the LLR scaling parameter based on channel conditions by using a neural network that outputs adaptive scaling factors. Instead of fixed scaling, the system continuously adjusts the scaling parameter according to observed channel characteristics, resolving the contradiction between adaptability and complexity.

Inventive Principle:
Principle #35Parameter changes

2Measurement precision

If LLR scaling is optimized for specific channel conditions, then decoding accuracy improves, but performance degrades when channel conditions change

Engineering Contradiction:
ImproveLLR scaling accuracyVSAvoidperformance across varying conditions
Core Design Contradiction:
Measurement precisionVSAdaptability or versatility

Solution Approach 1:

The system performs preliminary training of the neural network model using historical LLR data and a posteriori decoding metrics before actual decoding operations. This pre-training phase prepares the model to quickly adapt to different channel conditions, ensuring both accuracy and versatility when real-time decoding occurs.

Inventive Principle:
Principle #10Preliminary action

Solution Approach 2:

The patent implements a feedback mechanism where a posteriori decoding metrics from actual decoding operations are used to update and refine the neural network model. This continuous feedback loop ensures the scaling accuracy improves over time while maintaining adaptability to changing channel conditions through iterative model refinement.

Inventive Principle:
Principle #23Feedback

3Adaptability or versatility

If machine learning model training is performed in real-time, then adaptability improves, but processing time increases during active decoding

Engineering Contradiction:
Improvereal-time adaptation capabilityVSAvoiddecoding processing time
Core Design Contradiction:
Adaptability or versatilityVSLoss of time

Solution Approach 1:

The system performs model training periodically during idle periods rather than continuously during active decoding. Unused LDPC decoder accelerators are utilized during non-peak times to train the neural network model, allowing real-time adaptation capability to be maintained without increasing processing time during actual decoding operations.

Inventive Principle:
Principle #19Periodic action

Solution Approach 2:

The patent introduces an intermediary training process that uses unused decoder resources and historical data to prepare the model in advance. This intermediary training phase separates the computationally intensive model updates from the time-critical decoding operations, eliminating the trade-off between adaptability and processing speed.

Inventive Principle:
Principle #24Intermediary (Mediator)

Data Source

PatentEP4651384A1Self-adapting low-density parity check (LDPC) decoder architecture
Publication Date: 2025.11.19 INTEL CORP
  • EP4651384A1 patent drawingFigure 1
  • EP4651384A1 patent drawingFigure 2
  • EP4651384A1 patent drawingFigure 3

AI summary

A low-density parity check (LDPC) decoder architecture is provided for a self-adapting an LDPC control input that is used to successfully decode corresponding received code blocks. The architecture enables a machine-learning based process that facilitates learning of optimal LDPC control inputs, such as log-likelihood (LLR) terms scaling, that is required for a given channel condition and deployment scenario. This is achieved by gathering, as a background process, a posteriori decoding metrics via unused LDPC decoder accelerators, which function to process the labelled LLR data sets for different LDPC control input values. Then, such optimal LDPC control input estimates may be applied to the real-time LDPC decoding based on previous learning across UE/channel conditions.