Research On Blind Identification Algorithm Of Non-censored Polarization Code Parameters Based On Parity Check

Channel coding technique is mainly used in digital communication and is a data processing technique used to increase the reliability and improve the channel capacity during transmission.Polarization codes are the only coding type that has been proven to achieve the Shannon limit transmission rate at code lengths close to infinity,and blind identification of channel coding has been a very popular research direction in recent years to obtain valid information based on intercepted code stream data to complete the identification of coding parameters in cognitive radio and noncooperative communication scenarios.Therefore,it is very important to propose an algorithm for polarized code blind recognition research.The main contents and innovations of this paper are summarized as follows.(1)Recognition algorithm is divided into closed set recognition and open set recognition,closed set recognition is to make the user device with code length,code rate,freeze bit bit these a priori knowledge for semi-blind recognition,but these a priori knowledge is difficult to be satisfied in the field of non-communication cooperation;open set recognition method is mostly to build the candidate set or build the pairwise space of generation matrix,but the large-scale operation will increase the complexity and affect the algorithm performance.To address the above problems,from the perspective of reducing complexity and improving recognition rate,a blind recognition algorithm for non-censored polarization code parameters based on parity check is proposed.Since the third-party interception stream can only obtain the output soft information and the hard judgment sequence corresponding to the soft information,the intercepted hard judgment sequence is segmented according to the integer multiple of 2,and if it is a polarization code,there is a corresponding generation matrix.By mathematical induction,under the condition of error-free code,the row vector obtained by adding a section of hard judgment sequence and the column vector of the generation matrix by modulo 2 is the original data bit sequence,then each section of hard judgment sequence and the generation matrix can be multiplied to obtain a matrix,whose all "0" columns correspond to the frozen bits in the polarization code,then the original data bits are The probability of freezing bits can be converted into the probability that the base of the element "1" in each sequence code word is even,and the parity check relationship can be obtained,and the blind identification result can be obtained by verifying whether this relationship is consistent with the data distribution of the original data bits.(2)Likelihood Difference(LD)is introduced as the accuracy index of parity check results,and the recognition performance of this method becomes less complex while improving compared to LogLikelihood Ratio(LLR).The statistical features are obtained by calibrating each segment with the average LD value to set the optimal judgment threshold according to the receiver’s requirement for false alarm probability or missed alarm probability.The code length is traversed and the correct code length and code rate are obtained based on the size relationship between the average LD value and the judgment threshold to distinguish the type of intercepted sequence and the information bits.According to the simulation,this algorithm performs significantly better than the algorithm proposed in the literature [42],and still has good recognition ability under the low Signal Noise Ratio(SNR)Additive White Gaussian Noise(AWGN)channel.(3)The existing polarization code recognition technology almost does not consider the situation when the freeze bit does not take 0,in the non-communication cooperation environment,the value of the freeze bit is not directly translated into 0,so consider the case when the freeze bit takes 1,then the above matrix of all "0" column or all "1 The freeze bits correspond to all "0" columns or all "1"columns in the above matrix,and the even "1" and odd "1" cases in the segmented hard judgment sequence are considered separately to obtain two judgment thresholds,and then the average LD value and the judgment threshold are compared between the two.According to the simulation results,the values of the frozen bits can be effectively distinguished and the parameter values for blind recognition can be obtained.Although the recognition capability does not differ much from that of the literature [39],the complexity is greatly reduced.