Research On Blind Recognition Of Convolutional Codes

Since the birth of Shannon’s theorem,communication transmission technology has continuously developed and progressed to approach Shannon’s limit.Among them,in order to overcome the noise interference encountered during channel transmission,modern communication systems adopt various forms of high-performance channel coding technology.As an important member of channel coding,convolutional codes have been widely used in modern civil communications and military countermeasure communications.For non-cooperative communication scenarios,such as communication confrontation scenarios or adaptive modulation and coding application scenarios,the channel coding parameters need to be blindly identified before corresponding channel decoding operations can be performed.This thesis focuses on the blind recognition method of convolutional codes.The main work and contributions are as follows:Firstly,this thesis studies the blind recognition algorithm of convolutional codes under the condition of low bit error rate,and realizes the blind recognition algorithm based on the reconstruction analysis matrix method.The algorithm can identify the basic convolutional codes with less received bits.Encoding parameters.In order to further reduce the complexity of the algorithm,this thesis proposes an improved C(n,1,m)convolutional code generation matrix recognition algorithm on this basis,The C(n,1,m)convolutional code is divided into N/2 C(2,1,m)convolutional codes,and then each group of convolutional codes is identified respectively,which can effectively reduce the complexity.Secondly,the blind recognition algorithm of convolutional codes is studied under the condition of high error rate.This thesis is based on the Walsh-Hadamard transform method to solve the error-containing equation,and then with the aid of the decision threshold,the basic parameters of the convolutional code can be accurately identified.Based on the improved grouping algorithm,the analysis matrix is truncated to optimize the amount of data required for single identification.The simulation results show that the proposed algorithm has lower complexity than existing algorithms under the premise of ensuring recognition performance.Lastly,this thesis studies the blind recognition algorithm of convolutional codes based on neural network.Traditional blind recognition algorithms based on statistics and matrix processing have high computational complexity and limited application scope.In order to solve this problem,this thesis proposes a blind recognition scheme based on LSMT neural network.By analyzing the similarities between the memory characteristics of the LSTM network and the retrospective characteristics of the convolutional code,a blind recognition model is established to realize the blind recognition of the basic parameters of the convolutional code and the generating matrix.The simulation results show that the algorithm achieves blind recognition of convolutional codes with low error codes,and at the same time has lower complexity than traditional algorithms.The three convolutional code blind recognition algorithms studied in this thesis are applicable to different scenarios,namely low error,high error,and low complexity.The proposed algorithm can effectively reduce the complexity while ensuring the recognition performance.And the algorithms have certain practical application value.