| In mobile communication system,channel coding technology is the key technology to ensure transmission reliability.Low density parity check(LDPC)codes have performance approaching Shannon limit and good hardware parallelism of decoding and are widely used in communication systems such as Wi MAX,Wi-Fi and 5th generation mobile communication.NMS(Normalized Min Sum)decoding algorithm is an important decoding algorithm of LDPC codes which can realize per-formance approaching Shannon limit under low complexity.However,there is still room for im-provement in the performance and complexity of the algorithm under the scenario of high reliability and low delay.This paper focuses on the low complexity decoding technology and application of LDPC codes.The main studies include NMS quantization decoding algorithm and parameter optimization of the al-gorithm,the post-processing quantization decoding algorithm of LDPC codes based on trapping sets,the three states quantization decoding algorithm of PB-LDPC(Protograph-Based LDPC)codes and the related technologies of multi-user MIMO(Multiple Input Multiple Output)uplink transmission of LDPC codes.The main research work and contributions include the following contents:This paper introduces NMS quantization decoding algorithm and the decoding platform based on single instruction multi data stream.The attenuation factorαand quantization factorηaffect the performance of NMS quantization decoding algorithm.Traversal algorithm can find the best factor combination but has high complexity.In order to optimize factor combination with low complexity,grid search algorithm is proposed and the algorithm is implemented in the decoding under specific signal-to-noise ratio in the waterfall area.The decoding simulations of(8000,4000)regular LDPC code and(22,12,96)NR-LDPC(New Radio LDPC)code show that the decoding performance in the waterfall area under the factor combination optimized by grid search algorithm can approach the performance under best factor combination.This paper analyzes the relationship between trapping sets and error floor of LDPC codes,and introduces the MS decoding algorithm based on threshold.The decoding simulation of(8000,4000)regular LDPC code shows that the algorithm reduces the error floor by about 1/2 order of magnitude.But the algorithm can’t reduce the error floor of NR-LDPC codes’decoding.NR-LDPC codes defined by BG2 have short code length which are excellent solutions for channel coding in high reliability and low delay scenarios.In order to reduce the error floor of NR-LDPC codes’decoding,the post-processing quantized decoding algorithm of LDPC codes based on trapping sets is proposed in this paper.After NMS quantization decoding,the algorithm flips the possible error information bits of the codeword that is judged to be wrong by CRC(Cyclic Redundancy Check),based on the error and offset modes of variable and check nodes of trapping set clusters.The decoding simulations of NR-LDPC codes with different expansion factors(Z=384/144/96)and different rates(R=2/3/1/2/1/4)show that the post-processing quantization decoding algorithm can reduce the error floor by about one order of magnitude.This paper introduces the construction of PB-LDPC codes and the implement of the three state quantization decoding algorithm of PB-LDPC codes.The decoding simulations of(16,2,12×52),(12,2,13×64)and(10,2,16×62)PB-LDPC codes with different code rates and prototype graphs show that the waterfall performance of the algorithm can approach performance of 5-bit quantization NMS decoding with a loss of about 0.5d B but the decoding error floor performance is poor.In order to further reduce the complexity of the algorithm and lower the error floor,this paper proposes external information reliability quantization and quantization decoding post-processing algorithm.External information reliability parameters are important parameters of the algorithm which are different in the update of prototype graph check nodes with different iteration times.The decoding simulation of(10,2,16×62)PB-LDPC code shows that 4-bit quantization parameters can better balance the per-formance and complexity.The quantization decoding post-processing algorithm is used to optimize the error floor and simulation of(10,2,16×62)PB-LDPC code shows that the algorithm can reduce the error floor by about 1/2 orders of magnitude.This paper introduces the MIMO uplink transmission technology,the joint design of precoding and detection based on MMSE and the channel soft quantity calculation method of LDPC codes under high-order modulation.The multi-user MIMO uplink transmission system is designed using 16QAM modulation,(12,6,32)NR-LDPC codes and a variety of precoding solutions.The simulations show that the direct transmission of precoding matrix performs best in transmission performance and total system throughput.The precoding solution of using 2-bit quantization codebooks balances the uplink transmission performance and downlink transmission resources.6-user and 8-user MIMO uplink transmission systems can work normally. |
