Research On Blind Recognition For Cyclic Code Parameters

In order to improve the reliability of data transmission and decline in a communication channel interference effects on communication system, the channel coding technology is usually adopted to reduce the bit error rate and achieve stable communication. With the widely application of the channel coding technology, the recognition problem of channel coding is becoming more and more significant. The purpose of channel coding recognition is to estimate the coding system and code parameters of the intercepted code sequence with just a few or even without prior information, so as to extract the original information sequence. So the blind identification of channel coding technology has important theoretical significance and practical application value in the actual information interception, the collaboration communication and intelligent communication and other fields. This paper researches on blind identification problem for cyclic code parameters.First of all, this paper introduces the theoretical knowledge of channel encoding, the mathematical model of encoding recognition and identification parameters. It lays the foundation for the blind identification algorithm of the following chapters. Secondly, summarize the recognition methods for cyclic code at present, and analysis the advantages and disadvantages, adapting object of every one. Then, according to the feature similarity measure function which appears in data mining, the code length and start point are identified by the characteristic of the large difference between the actual sequence and random sequence; Finally, according to the relationship between the code word polynomial and the generator polynomial, the decision threshold is set to solve the generation matrix, and the semi blind identification of cyclic codes is completed. The algorithm is easy to understand, the Simulation results show that it can exceed 90% about the recognition rate with 0.01 BER. Inspired by the above algorithm, the code length and starting point are identified by the recognition method based on the fusion of a similarity measuring function in data mining and Spearman Rank correlation coefficient in statistics according to improved theoretical and experimental simulation. Through the method based on the Galois field Fourier transform, the generator matrix is solved. Therefore, the whole blind recognition of cyclic code is finally realized. Finally, researching on the special cyclic code, that is to say the BCH codes: the code length are identified by the recognition method based on the coefficient of variation that is proposed by the most notable differences whose order of the largest common divisor probability distribution between the actual sequence and random sequence, by computing the order between the code word and its circular shift code word. On this basis, the starting points are identified by constructing the distribution probability. Then, The generator matrix is solved through the method of getting the maximum of the order distribution. The blind recognition of primitive BCH codes is finally realized. The method has less computation as well better error-tolerance. And the effect is obvious in the short-code recognition, in the BER of 0.02 conditions.