Research On High Performance Variable Factors Min Sum Based LDPC Decoding Algorithm

In modern communication systems,improving the quality of data transmission has always been the goal of academic and industrial world and it's the source of new technology.For example,in the digital TV transmission system,the 4K TV,8K TV and so on not only put forward higher requirements for the data transmission quality,but also put forward more demand for the spectrum resources of data transmission.Based on the Shannon theorem,the requirement of data transmission quality and spectrum resource can be equivalent to reduce the SNR threshold of communication system on the premise of given spectrum bandwidth and data transmission rate.In the layered architecture of modern communication systems,the physical layer has always been the key to reduce the threshold of the communication system.Furthermore,the channel coding technique used in the physical layer is one of key technologies to improve the system performance.Low density parity check code(LDPC)is one of the widely used channel coding techniques in the existing communication systems.To achieve lower system threshold and higher data transmission quality,this paper focuses on improving the performance of LDPC decoding algorithm and does some work on the following three aspects.First,base on maximum a posteriori estimation,a general optimization criterion of variable modifying factors call MAP-VFP is derived which is applicable for various modified min sum algorithms and various LDPC decoding architectures.Second,variable scaling factor decoding algorithm in flooding structure(FS)is extended to horizontal shuffled structure(HSS),and base on MAP-VFP,the method of optimizing variable scaling factors is derived.The simulation results in this paper show that under HSS,for the LDPC code from DVB-T2 standard with code length 64800 and code rate 1/2,when bit error rate is 1e-7,variable scaling factor(VNMS)decoding algorithm has a advantage of 0.13 dB compared with fixed scaling factor(NMS)decoding algorithm.And when signale noise ratio is 0.9dB,the everage iteration time of VNMS algorithm is 32.5% less than NMS algorithm.Last,the nonlinear functional modifying factors min sum(NLVMFMS)decoding algorithm and its simplified(S-NLVMFMS)algorithm are proposed.Based on MAP-VFP,NLVMFMS and S-NLVMFMS algorithm are used in FS,HSS and vertical shuffled structure(VSS)and the simulation results show that NLVMFMS and S-NLVMFMS algorithm are better then the VNMS algorithm not only in the decoding performance but also the decoding convergence speed.For example,under VSS,for the LDPC code from ATSC 3.0 standard with code length 64800 and code rate 10/15,when bit error rate is 1e-7,NLVMFMS algorithm have a advantage of 0.04 dB compared to VNMS algorithm while S-NLVMFMS algorithm have a advantage of 0.02 dB compared to VNMS algorithm.Through these three aspects of work,a set of high-performance decoding algorithms based on min sum algorithm which are suitable for all kinds of decoding architectures are formed.