Multi-code-rate and multi-code-length ldpc code decoding method based on simd instruction set

Abstract

The present invention provides a multi-code-rate multi-code-length LDPC code decoding method based on the SIMD instruction set, combined with the characteristics of the LDPC code base matrix of the quasi-cyclic structure, realizing online statistics of check matrix information through external configuration files, and adopting fixed-point layered decoding In the scheme, the decoder constructs a specific check node computing unit for different row weights, selects the check node computing unit according to different row weights, and adopts a loop expansion method inside the check node computing unit. The present invention realizes the online statistics of check matrix information, reduces the storage capacity of the LDPC code decoder with multiple code rates and code lengths; compared with the existing algorithm, it eliminates the dependence on matrix statistical information, and reduces the need for decoder implementation. Complexity: Compared with the existing algorithm, there is no speed loss, and at the same time, the module of online calculation can be modified, which has universality.

Description

technical field [0001] The invention belongs to the technical field of channel coding, and specifically relates to a decoding method, more specifically, relates to a multi-code-rate and multi-code-length LDPC code decoding method based on a SIMD instruction set. Background technique [0002] The LDPC code is the Low-Density Parity-Check (LDPC) code, which was first proposed by Dr. Robert G. Gallager in 1963 as a linear block code with a sparse parity check matrix and was adopted by D.J.C.Mackay in 1995 and R.W. Neal re-presented. LDPC codes have good performance close to the Shannon limit, and have low decoding complexity, flexible structure, low delay, and high-speed parallel decoding. It has been a research hotspot in the field of channel coding for nearly two decades. At present, LDPC codes have been widely used in wireless communication (WiFi, WiMAX), deep space communication, optical fiber communication, satellite digital video and audio broadcasting and other fields. ...

AI Technical Summary

Problems solved by technology

[0006] The LDPC decoder implemented by pure software has attracted the attention of many researchers in recent years. However, the research in recent years has mainly focused on the flooding method of LDPC decoding and GPU implementation, which is not the most effective LDPC decoding. device implementation

The latest research (Reference 1: Gal B L, Jego C. High-Throughput Multi-Core LDPCDecoders Based on x86Processor [J]. IEEE Transactions on Parallel & Distributed Systems, 2016, 27(5): 1373-1386.) proposed a method based on A high-speed decoder for LDPC codes based on the X86 architecture. However, this method requires off-line statistical information about the parity check matrix, and the implementation complexity is high. It is difficult to apply it to a class of 5G-oriented LDPC codes that support multiple code rates and multiple code lengths.

[0007] Among the LDPC decoding algorithms that have been proposed so far, the layered decoding algorithm has the characteristics of fast convergence speed and less memory usage and has attracted extensive attention. Although the parallelism of the layered decoding algorithm is not high, it is suitable for X86 architecture processors. Software implementation; the current existing research shows that the logarithmic likelihood ratio sent to the LDPC code decoder adopts 6-bit quantization, and the decoder adopts 8-bit processing mode. Compared with fixed-point decoding and floating-point decoding, the performance loss can be ignored Excluding

Images