Study On Low Density Pairty Check Code For The Digital Broadcasting Communication Systems

Because of its good performance,the low density parity check(LDPC)code has been widely studied by many reaserchers.And it has been profoundly used in digital broadcasting communication systems.As broadcasting communication system continue to evolve,it uses some new technologies including LDPC code with Raptor-like structure and Non-uniform constellation(NUC).Meanwhile,the evolved digital broadcasting system also proposes some higher requirements for the physical layer technologies,such as wider coverage,higher transmission capacity,more robust transmission and lower receiver algorithm.However,when we try to satisfy the requirements of evolved digital broadcasting systems,we find out that the existing LDPC technologies still have some issues.These issues are mainly concentrated in the following four areas: 1)Raptor-like LDPC code can achieve better performance at low code rate.So we can use low code rate Raptor-like LDPC code to satisfy the wider coverage requirement.However,existing algorithms can not effectively and easily analyze the performance of Raptor-like LDPC code.2)The theoretical transmission capacity of BICM system can be increased by using NUC constellation.But the algorithms which can analyze the performance of LDPC code in the BICM system with NUC constellation have very high complexity,and they can not be used easily.3)The existing design method of structural LDPC code can be improved by using new analysis and construction algorithms.Therefore,we improve the design method,and the improved method can design LDPC codes with better performance.4)When we decode the Raptor-like LDPC code,existing iterative decoding method can be improved,and the improved iterative decoding method can reduce the decoding complexity.Then,we make a profound study on these four areas.And we propose some new algorithms or improved methods to improve those issues and satisfy the requirements.First,this thesis proposes two density evolution analysis algorithms based on two-type degree distribution,which can be used to analyze the performance of Raptor-like LDPC code.Raptor-like LDPC code can achieve good performance at low code rate because of its special structure.However,the traditional density evolution algorithms based on degree distribution and multi-edge type(MET)distribution both have some issues for the analysis of Raptor-like LDPC code.The algorithms based on degree distribution cannot analyze Raptor-like LDPC code correctly because they cannot describe the special structure.Meanwhile,although the algorithms based on MET distribution can analyze Raptorlike LDPC code correctly,the complexity is high,and it is very difficult to be used.So,we fully consider the special structure of Raptor-like LDPC code and propose the two-type degree distribution framework.This framework classifies the variable nodes and check nodes of Raptorlike LDPC code according to the special structure.Then,we use twotype degree distribution to describe Raptor-like LDPC code based on the classified nodes.Finally,based on this proposed two-type degree distribution,we propose the information transfer process during the decoding of Raptor-like LDPC code,and propose the density evolution algorithms.The proposed algorithms can effectively analyze the performance of QC-Raptor-like LDPC code.And it can reduce about 80%algorithm complexity.Second,this thesis proposes an analysis algorithm,which can analyze performance of LDPC code in the BICM systems with NUC technology.And this proposed algorithm can reduce the complexity by using mixture Gaussian approximation.NUC technology is applied by the evolved digital broadcasting communication system,because it can improve the maximum transmission capacity of BICM system.However,if the system wants to fully achieve the performance gain of NUC technology,the LDPC code should cooperate with the NUC technology.So the performance analysis of LDPC code in BICM system with NUC technology is very important.However,the traditional algorithm can only analyze the performance based on discrete density evolution in this condition.Because the discrete algorithm is a high dimension calculation,the algorithm complexity is very high.So,we analyze the distribution of soft information output by the NUC demodulation,and find out that we can use mixture Gaussian distribution to approximate these distribution of soft information.Meanwhile,most LDPC codes used in BICM system are quasi-cyclic(QC)LDPC code,and protograph can describe QC LDPC code appropriately.Therefore,we use the mixture Gaussian approximation distribution to derive the performance analysis algorithm based on the protograph.This algorithm can analyze the performance of LDPC code in the BICM system with NUC technology.During the analysis,this algorithm only calculates the means of the Gaussian components of the mixture Gaussian distribution.So,this algorithm is a low dimension calculation.Compared with the discrete analysis algorithm which is a high dimension calculation,this proposed algorithm can reduce the complexity.Next,we improve the traditional design method of structural LDPC code.The traditional design method of structural LDPC code is as follows.Firstly,we choose a LDPC structure.Then,we design the degree distribution or MET distribution.Finally,we construct the QC LDPC code based on the designed distribution.In this thesis,we improve this traditional design method.First,we find out that the design of MET distribution is very difficult because of the high complexity,when we design the QC-Raptor-like LDPC code.So,we use the design method based on the proposed two-type degree distribution to replace the design method based on the MET distribution.Next,we add the design based on protograph after the design based on degree distribution.Adding the design based on protograph can not only improve the performance of LDPC code,but also provide all the possible short girths for the QC LDPC construction.Finally,we also propose a QC LDPC construction algorithm based on the designed protograph.This proposed construction algorithm can obtain all the locations of possible short girths from the designed protograph,and it constructs the QC LPDC codes by avoiding constructing these possible short girths.The simulation results show that the improved design method can effectively design structural LDPC code with good performance.In many different simulation conditions,the improved design method can provide0.05?0.1dB performance gain.Finally,this thesis also proposes a low complexity iterative decoding method based on Raptor-like LDPC code.As the number of handheld mobile terminals increases,the evolved digital broadcasting systems also wants to be used in these terminals.However,these terminals usually require the decoding method with low complexity.Therefore,this thesis studies how to reduce the complexity of decoding method based on Raptor-like LDPC code.We analyze the iterative decoding process of Raptor-like LDPC code with good performance,by using the proposed algorithm based on two-type degree distribution.According to the analysis,we find out an interesting phenomenon,which is that some soft information is approximated to zero during the early decoding iteration.Then,we propose an improved iterative decoding method of Raptor-like LDPC code based on this interesting phenomenon.This improved iterative decoding method skips the calculations of the soft information which can be approximated to zero during the early decoding iteration.And this improved method can reduce the decoding complexity without performance sacrifice.According to the simulation results,this improved decoding method can reduce about 10%complexity on average.Meanwhile,this improved method is useful for several iterative decoding algorithms,such as belief propagation(BP)algorithm,normalized min sum(NMS)algorithm and offset min sum(OMS)algorithm.