Research And Design Of LDPC Code With Fast Encoding Structure

LDPC code is a type of channel error correcting code which is closest to Shannonlimit currently, and is also a focus in encoding field. Due to its excellent performance,LDPC code gradually received widely used in communication systems and storage devices.And LDPC code with structure of easy to hardware implementation becoming an importantresearch subject.This dissertation firstly have in-depth study on the encoding and decoding algorithmof LDPC code, and give a detailed exposition for the principle of hard decision decoding,BP decoding, Log_BP decoding and min_sum decoding algorithm. On the basis of thedecoding algorithm, this dissertation analysis and research on the impact factor for LDPCcode’s performance, such as degree distribution, girths, code length, code rate and thenumber of iterations. Subsequently, this dissertation introduce to three methods for degreedistribution optimization: density evolution, Gaussian approximation and EXIT chart, andcompare these three algorithms by the noise threshold and the complexity. Then thisdissertation using DE algorithm and EXIT chart method to search the degree distribution ofirregular LDPC codes and found that the degree distribution based on the DE-EXIT chartmethod has a high noise threshold.on the basis of the previous work, this dissertation give a complete process of LDPCcodes with fast encoding structure: First, determine the size of basis matrix, according tothis scale and the good degree distribution search by DE-EXIT chart method, then use thePEG and ACE algorithm to construct the base matrix, and finally construct the final checkmatrix through fill the base matrix by circulation unit matrix. Through performancesimulation, the codeword designed by this dissertation has about0.3dB better thanIEEE802.16e standard LDPC codes in the condition of corde length is2304, code rate is1/2, frame is3000and BER is10-5.