Polar Encoding Matrix Layout for Lower Decoding Complexity
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Current polar coding methods face high encoding/decoding complexity, especially with large code lengths, leading to poor performance.
Innovation Solution
The proposed solution involves generating a second generator matrix based on a first generator matrix, which includes sub-blocks distributed according to a preset position relationship. This allows for polar encoding of K to-be-encoded bits, equivalent to encoding multiple short codes and coupling them, thereby reducing complexity.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Length of stationary object
If polar code encoding/decoding is performed with large code length, then code length increases, but encoding/decoding complexity becomes very high
Solution Approach 1:
The patent divides the large code length N into multiple smaller code lengths n by constructing a generator matrix G_N composed of multiple sub-matrices G_n arranged in a specific pattern. This segmentation allows the encoding process to be broken down into multiple smaller encoding operations rather than one large operation, thereby reducing the overall encoding/decoding complexity while achieving the desired large code length.
2Device complexity
If polar code encoding/decoding complexity is reduced, then encoding/decoding performance improves, but code length capability is limited
Solution Approach 1:
The patent employs a nested structure where multiple generator matrices of smaller code length n are embedded within a larger generator matrix of code length N. The generator matrix G_N is constructed by arranging multiple G_n matrices in a specific pattern with zero matrices, creating a hierarchical structure that enables the system to achieve large code length N while performing only smaller-scale encoding operations, thus maintaining low complexity.
3Reliability
If traditional polar code methods are used, then encoding/decoding can be performed, but performance is poor with large code lengths
Solution Approach 1:
The patent applies local quality by designing a specific structure for the generator matrix where different regions serve different purposes. The generator matrix G_N is constructed with non-zero sub-matrices G_n positioned at specific locations and zero matrices at other locations, creating a structured sparsity pattern. This local structural design optimizes the encoding performance for large code lengths by ensuring that information is distributed across multiple reliable sub-codes rather than being concentrated in a single large code.
Data Source
AI summary
An encoding method and apparatus, a decoding method and apparatus, and a device are provided. The encoding method includes: obtaining K to-be-encoded bits (S301), where K is a positive integer; determining a first generator matrix, where the first generator matrix includes at least two sub-blocks distributed based on a preset position relationship, and the sub-block includes a plurality of first generator matrix cores (S302); generating a second generator matrix based on the first generator matrix, where the second generator matrix includes T sub-blocks, and a position relationship between two adjacent sub-blocks of the T sub-blocks is determined based on the preset position relationship (S303), where T is a positive integer; and polar encoding the K to-be-encoded bits based on the second generator matrix (S304), to obtain encoded bits. This can reduce encoding/decoding complexity.


