Structured LDPC Base Matrix Layout for High-Speed Layered Decoding
Find Innovative SolutionsGenerate Solutions
Solution Overview
Problem
Conventional LDPC decoders/encoders suffer from low efficiency due to the need for processing one row of the base matrix before starting the next stage in layered decoding, leading to reduced efficiency, especially in high bit rate scenarios, and there is no feasible method to reduce waiting time for varying code lengths.
Innovation Solution
The use of a base matrix with K0 up-and-down adjacent pairs, where K0=K1+K2, allows for the same base matrix to be used across different code lengths, reducing storage needs and enabling high pipeline speed LDPC encoding and decoding by eliminating the need for a cyclic shift inverse network, thereby increasing efficiency.
Engineering Contradictions & Design Principles
Engineering Contradiction Analysis
1Speed
If conventional layered decoding is used where one row of the base matrix is processed before starting the next stage, then decoding can be performed systematically, but processing speed is reduced and waiting time increases
Solution Approach 1:
The base matrix is segmented into multiple groups that can be processed in parallel. Instead of processing one row at a time sequentially, the decoding process divides the base matrix into segments that enable simultaneous processing of multiple rows, thereby increasing throughput and reducing waiting time while maintaining systematic decoding.
Solution Approach 2:
The decoding architecture transitions from a static sequential processing approach to a dynamic parallel processing approach. By reorganizing the processing stages to allow multiple rows to be handled concurrently through dynamic resource allocation and pipeline parallelism, the system achieves higher processing speed without sacrificing decoding accuracy.
2Reliability
If different base matrices are used for different code lengths, then decoding accuracy can be optimized for each code length, but storage requirements and device complexity increase
Solution Approach 1:
A single base matrix is designed to serve multiple code lengths through a universal structure. The base matrix incorporates a pattern that can be adaptively applied to different code lengths without requiring separate matrices for each length, thereby reducing storage requirements and device complexity while maintaining optimized decoding accuracy across various code lengths.
Solution Approach 2:
Instead of changing the base matrix structure for different code lengths, the system maintains a fixed base matrix and adjusts processing parameters such as the number of iterations, parallel processing degree, or extraction patterns to optimize decoding accuracy for each specific code length. This parameter-based adaptation eliminates the need for multiple stored matrices.
Data Source
Figure 1~3
Figure 4~6
Figure 7~9
AI summary
An encoding method, decoding method, encoding device and decoding device for structured LDPC codes. The method includes: determining a basic matrix used for encoding, which includes K0 up-and-down adjacent pairs; and according to the basic matrix and an expansion factor corresponding to the basic matrix, performing an LDPC encoding operation of obtaining a codeword of Nb×z bits according to source data of (Nb-Mb)×z bits, herein z is the expansion factor, and z is a positive integer which is greater than or equal to 1. The provided technical solution is applicable to the encoding and decoding of the structured LDPC, thereby realizing the encoding and decoding of LDPC at the high pipeline speed.