The Parameter Estimation Algorithm Of Non-punctured Systematic Convolutional Code

With the wide application of channel encoding in digital communication and the rapid development of communication reconnaissance and anti-reconnaissance technology,the analysis of the channel encoding has evolved from the traditional cooperative manner to the non-cooperative one.Blind recognition technology of channel encoding schemes can get the channel encoding parameters with the absence of priori information,which lay the foundation for the subsequent information processing.Therefore,the blind recognition of channel encoding is of great pratical significance,which can be widely used in intelligent communication,network confrontation,and communication reconnaissance fields.This thesis mainly focuses on the investigation of parameter estimation algorithms for non-punctured systematic convolutional code.Firstly,the research background and significance of this topic is given,which is followed by the status quo of channel coding recognition technology.Secondly,as for the non-punctured convolutional codes,the working principle and characteristics of existing convolutional code recognition algorithms are investigated,which leads to a multidimensional feature serving as a basis of parameters estimation for non-punctured convolutional encoding.The performance of these algorithms is also analyzed via numerical simulations.Thirdly,this thesis makes concrete analysis of the recognition principle,and recognition performance via simulation for the convolutional code recognition method based on Euclidean method.To overcome the defects that the existing algorithm can only identify 1/2rate convolutional codes and code words synchronization must be ensured before recognition,an improved algorithm is proposed which guarantees code words synchronization through synchronous searching based on Viterbi decoding.By solving the greatest common factor of many instead of two code words polynomials,it can accommodate the recognition of k/n rate convolutional codes.Moreover,the amount of computation is reduced fromO?N ²?to O?N?.Finally,this thesis carries out detailed analysis of the recognition principle,and performance of the convolutional code recognition method based on Gauss Jordan Elimination.To overcome the defects of the poor adaptability of error code of the abovementional algorithm and the require of the prior knowledge on starting point of code words before identifying the check vector,an improved algorithm is proposed,which can get the starting point of code words by matrix analysis and ensure the check vector being recognizable by starting point correction,and reduces the required prior information.Meanwhile,by traversing non-zero correlation columns,the maximum probability detection is introduced to improve the adaptability of error code.MATLAB simulation results demonstrate that the average recognition rate is more than 99%when bit error rate is not greater than3×10^-4.