The Research Of LPDC Codes Blind Recognition

Channel coding recognition means that the receiver obtains the corresponding coding parameters after analyzing and calculating the received message sequence,which is the foundation of recovering the message sequence.Besides,it is widely used in the military communication and mobile communications.Thus,the researches of channel coding recognition have the important theoretical significance and research value.Low-Density Parity-Check code is a kind of block codes.It is widely used in deep space communication,satellite communication,optical communication and other military or civil fields due to its excellent characteristics such as closing to the Shannon limitation,simply decoding and the ability of strong error correction.At present,the researches of LDPC codes recognition compared with other types of block codes are few in China and on aboard.Thus,it is extremely important to develop researches of LDPC codes recognition.Restoring message sequences is the first goal of LDPC codes recognition,and the key content is the recognition of check matrix.So,how to improve the recognition ratio of check matrix has become the core problem in coding recognition.In addition,the recognition technology of LDPC codes mainly utilizes the log-likelihood Ratio(LLR)function to calculate the possibility of check codeword relationship.The formula involves the calculation of hyperbolic tangent function.But on account of the computational complexity of hyperbolic tangent function being high,the approximate formula can only be used to replace the calculation in practice.Therefore,the information in the channel cannot be fully utilized and recognition performance is worse either.This paper focuses on studying the influence of cosine test function instead of LLR on the recognition performance of check matrix and the decoding gain under the condition of closed and open set recognition.The main contents are summarized as follows:First,the closed set recognition model of LDPC codes is constructed.According to the check relation between the codewords and the check vectors,the corresponding expression is obtained based on the cosine test function.Then,the characteristics of probability distribution under two different hypothesis are analyzed,and the expressions of mean and variance are deduced,as well as the average statistics are carried out.The set of check matrix is known and the formula of recognition rate is derived according to the criterion of the maximum mean index of cosine test function.Finally,the simulation is obtained,and the experimental results verify the correctness of the theory,as well as solve the problem of channel information loss caused by approximate substitution.Secondly,under the condition of opened set recognition of LDPC codes,the amount of check vectors are reduced by constructing the pre-search space.In the pre-search space,the decision threshold formula is presented according to the characteristic of cosine test function and the preliminary decision is made.Then a comparator is designed to analyze the vector to be identified by the ratio of the variance under the two hypothesis,and finally all required check vectors are found.In order to improve the performance of the algorithm,the received data is iterated with the validated vector,and the verified matrix is finally recovered.Compared with the LDPC codes check vectors recognition algorithm based on soft decision,the proposed algorithm can find more check vectors,improve the average recognition rate by about 12%,and improve the decoding gain by 1d B.