Research On Blind Reconstruction Of Large Length Linear Block Code

Modern communication systems often adopt channel coding technology as one of the solutions for the validity and reliability of data transmission.In the field of non-cooperative and intelligent communication,it is a very important and fundamental problem to identify the parameters of error-control codes in noisy environment,that is channel code blind recognition.For LDPC codes and BCH codes,when the code length is larger and the incorrect bits is more,the computational complexity of the recognition algorithm is higher.the identification work is not able to be completed within a limited time for existing algorithm.This thesis focuses on the blind recognition algorithm of two types of linear block codes,quasi-cyclic LDPC codes and BC H codes in the case of large code length.The main tasks include:1)Introducing the basic mathematics,related important concepts for channel code and several important algorithms of high-order finite fields,that will be used in long code recognition.2)Based on the search algorithm of dual codewords,a blind recognition algorithm of code length and codeword deviation for rapid identification of long block codes is proposed.This algorithm uses the ISD(Information Set Decoding)algorithm to quickly identify dual codewords,compared with the traditional finite field method,it can quickly identify the code length and code deviation of the linear block code with larger code length.In the case that the error code does not exceed 10^-3,the recognition accuracy of the code length and code deviation can reach 100%.The average recognition time is only half of the traditional algorithm,and it has better error resistance performance.3)Giving the finite field representation of the blind recognition model of quasi-cyclic LDPC codes based on the algebraic theory of finite field and the aforementioned block code identification method.with this model,a quasi-cyclic LDPC long code blind recognition method based on the finite field ISD algorithm is proposed.This method uses the finite-field representation of the quasi-cyclic LDPC code,and uses the divide-and-conquer technique to greatly reduce the quantity of calculation for blind recognition of LDPC codes with the shape of the LDPC codes in the IEEE 802.16^{T M}protocol.for LDPC codes with a code length more than 2000,the code length and parity matrix can still be quickly and completely identified under the condition that the bit error rate does not exceed 0.00049.4)A large code length BC H code blind recognition algorithm based on dual codeword space is proposed.This method is mainly used for blind recognition of primitive BCH codes with a length larger than 511.Using the aforementioned ISD algorithm and the characteristics of the BCH code system parity matrix,it can effectively and quickly identify the relevant parameters of the primitive BC H code when the error code not exceeding 0.00049 and the code length not exceeding 1023.