| With the increasing demand for people's mobile communication relia-bility,especially after LDPC codes have been identified as the channel cod-ing schemes for important communication standards such as 5G and ATSC3.0,the research of LDPC code encoding and decoding have become increas-ingly urgent.As the application scenarios continue to be enriched,new re-quirements are also imposed on the encoding and decoding of LDPC codes.For example,in some scenarios,to ensure normal operation of all receivers,traditionally,the broadcaster may transmit signals in an excessive mode to ensure that every user receives a sufficient signal,which may lead to an over-engineered network.In addition,traditional broadcasting and mobile com-munication are increasingly converging.For example,The En TV(Enhanced TV)technology for broadcasting has been added to 5G standard.Above in-formation indicates that in the future,broadcasting signals will soon enter mobile phones and Io T devices.However,in practice,those devices are of-ten quite power limited.And the traditional broadcasting signals and their receiving equipments have excessive pursuit of performance.As a result,when broadcasting signals enter to power-limited deceives,there are some problems of excessive performance and high complexity.Therefore,it is valuable to reduce the complexity of the receiver with an acceptable perfor-mance when SNR is sufficient.This paper focused on the Raptor-like LDPC code and its decoding al-gorithm.The parity check matrix of the QC-Raptor-like LDPC code has many columns with degree of 1,which makes its decoding complexity higher.Firstly,the convergence of QC-Raptor-like LDPC codes and the boundedness of information transfer are proved in theory.The concept of skipped check nodes is proposed.Based on the skipped check nodes,this pa-per proposed a low-complexity decoding algorithm suitable for QC-Raptor-like LDPC codes,and correspondingly proposed an algorithm based on den-sity evolution theory to decide the skipped check nodes.Simulation results show that under different SNR conditions,the bit error rate can be guaran-teed at 9×10-7?9×10-6,while reducing the complexity by more than20%.However,directly ignoring the update operation of the skipped check nodes will cause a performance loss in the actual process,this paper used the external information of the skipped check nodes to participate in iter-ative decoding,which is effective to educe the loss of performance.The simulation results show that using external information after 2 iterations can achieve better performance.Then this paper analyzes the reasons for the performance loss of the normalize min sum decoding algorithm compared to the sum product decoding algorithm.According to the characteristics of the proposed decoding algorithm,an algorithm to optimize correction factor based on generalized mutual information criterion based on dynamic degree distribution is proposed,which made a great progress for the performance of proposed decoding algorithm.When the SNR is 4.85d B,the bit error rate performance can be improved from 1.4×10-6to 7.7×10-9. |
PDF Full Text Request