Research On Extended Min-sum Decoding Algorithm Of Non-binary LDPC Code

Digital communication has been widely used in all aspects of daily life,including mobile phones,WIFI,cable television.Among the related technologies of digital communication,channel coding,also named error control coding,is an indispensable part of all modern communication systems.Its main function is to ensure that the sink can recover the information sent by source accurately,overcome transmission errors caused by interference and fading in the mobile communication and enhance the ability of data to withstand various interferences when transmitting in the channel.Low Density Parity Check(LDPC)code is a hot spot in the field of channel coding in recent years,which has strong error correction performance and high-speed decoding potential when decoding.Compared with binary LDPC code,non-binary LDPC code shows better error correction performance.When continuous errors occur,non-binary LDPC code can obtain good error correction performance without interleaving deeply,which makes non-binary LDPC code better meet the needs of future mobile communication system.As one of the most practical algorithms in non-binary LDPC decoding,extended min-sum algorithm(EMS)algorithm has a rather low computational complexity.With the development of hardware,decoding computation is no longer a severe problem,and research has turned to improve the error correction performance of EMS.Further improving the error correction performance of EMS and retaining its advantages in decoding calculation as much as possible is one of the key problems in the research of LDPC code.Meanwhile,it also affects communication system research in the future.In this paper,we study the EMS algorithm from the aspects of check node update and variable node scheduling,and propose several methods to improve the error correction performance of EMS decoding algorithm and keep the advantages of decoding computation,which provides theoretical support for the research project.First of all,we introduce the genetic algorithm,and propose G-EMS algorithm by organizing the idea of combining genetic thought with non-binary LDPC code.The core point is to propose two constraints based on the reliability to filter reliable variable nodes in the decoding process.When the decoding reaches k iterations,reliable variable nodes are selected and the messages are optimized,which makes G-EMS algorithm improve the decoding performance on the basis of EMS algorithm.Then,based on the idea of reliability,we develop a dynamic scheduling strategy based on reliability of check nodes,namely CRS strategy.According to the reliability of check nodes,this strategy decides whether it is to be updated,and some variable nodes can no longer be updated.In addition,the message of some variable nodes is optimized according to the result of real-time scheduling.Compared with the usual flooding mechanism and LBP scheduling,this scheduling strategy improves the performance of error correction without obvious increase in decoding computation by selecting some nodes to decode and optimize the message.Finally,based on the CRS strategy,we further study and propose a reliable message correction algorithm based on hard information,which is to solve the problem of convergence speed reduction caused by factor correction and offset correction.Aiming at sub-reliable check nodes in CRS strategy,the proposed method optimizes the sub-reliable check nodes selected to improve the convergence rate reduction caused by the conventional method.Because this approach is based on CRS strategy,error correction performance can be further improved without further increasing computational complexity when used simultaneously.