Deep Polar Encoding Structure for Short-Block Decoding Accuracy

Resolve Bottlenecks,
Find Innovative Solutions
Generate Solutions

Solution Overview

Problem

Existing encoding methods using polar codes with a single layer result in decoding errors for short-length information bits, leading to inefficiencies.

Innovation Solution

Implementing deep polar encoding using polar kernel matrices connected to multiple layers, with varying matrix sizes and transpose relationships, and employing backpropagation-based deep polar decoding to enhance error correction.

Engineering Contradictions & Design Principles

VSEngineering Contradiction Analysis

1Reliability

If a single-layer polar code encoding method is used, then the device complexity is low, but decoding errors occur for short-length information bits

Engineering Contradiction:
Improvedecoding accuracyVSAvoidencoding structure complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

The encoding device is divided into multiple layers (first layer, second layer, third layer, etc.), each layer containing polar kernel matrices. This segmentation allows the system to process short-length information bits through multiple transformation stages, improving decoding accuracy while maintaining manageable complexity at each individual layer.

Inventive Principle:
Principle #1Segmentation

Solution Approach 2:

Multiple polar kernel matrices are nested within layers, and multiple layers are nested within the encoding device. Each layer processes the output of the previous layer, creating a nested structure where G1 processes information bits, G2 processes intermediate results, G3 processes further transformed data, and so on, enabling progressive error correction.

Inventive Principle:
Principle #7Nested doll (Nesting)

2Reliability

If deep polar encoding with multiple layers is implemented, then decoding errors for short-length information bits are reduced, but decoding complexity increases

Engineering Contradiction:
Improvedecoding accuracyVSAvoiddecoding structure complexity
Core Design Contradiction:
ReliabilityVSDevice complexity

Solution Approach 1:

Each layer is designed with specific local characteristics - the first layer uses polar kernel matrices for initial transformation, the second layer applies different polar kernel matrices for intermediate transformation, and subsequent layers continue with specialized transformations. This local optimization at each layer improves overall decoding accuracy while keeping individual layer complexities manageable.

Inventive Principle:
Principle #3Local quality

Solution Approach 2:

The decoding device inverts the encoding structure by using inverse matrices of the polar kernel matrices in reverse order. The decoding layers process data from the last layer back to the first layer, applying G3^(-1), then G2^(-1), then G1^(-1) to reconstruct the original information bits, enabling accurate recovery without proportionally increasing decoding complexity.

Inventive Principle:
Principle #13The other way round (Inversion)

Data Source

PatentEP4633044A1Enhanced depth polarization-based encoding and decoding methods, and apparatuses therefor
Publication Date: 2025.10.15 LEE NAM YOON
  • EP4633044A1 patent drawingFigure 1
  • EP4633044A1 patent drawingFigure 2
  • EP4633044A1 patent drawingFigure 3A

AI summary

Disclosed are a method for encoding and decoding based on improved deep polarization and an apparatus thereof. A code encoding method performed by an encoding device includes determining a first channel group connected to a lower encoding matrix and a second channel group connected to an upper encoding matrix and the lower encoding matrix among a plurality of polarized channels based on a weight of each of the polarized channels, the weight of each of the polarized channels being defined according to the number of specific binary data on a bit generated by each of the channels being polarized; and generating codewords by mapping information bits to the upper encoding matrix and the lower encoding matrix and assigning the generated codewords to the first channel group and the second channel group.