Blind Recognition Of Channel Coding Parameters In Non-cooperative Communication

The blind recognition technology of channel coding parameters is an important step in the process of information interception and analysis in non-cooperative communication.It means that unauthorized receivers can accurately identify the coding parameters based on the intercepted codeword sequence,and then use the relevant parameters for decoding.It is of great significance in the fields of military reconnaissance,intelligent communications,and information countermeasures.Firstly,RS codes,convolutional codes,Turbo codes,and Polar codes are selected as the representative coding types.The key parameters of each code are summarized,and the traditional recognition methods are simulated.The scope of application of each method and the recognition accuracy rate under different bit error rates are compared.Secondly,an improved CNN-based coding parameter recognition method is proposed,which uses traditional methods in coding recognition for feature extraction,and then constructs a convolutional neural network structure to convert the blind recognition of coding parameters into a one-dimensional sequence of fully supervised classification problem.The simulation results show that the proposed method performs well in various coding parameter blind recognition problems.This method also provides a feasible identification scheme for difficult problems that cannot be solved by traditional methods,such as the generation matrix of nonsystematic convolutional codes,the structure of Turbo code component encoders,and the distribution of frozen bits of Polar codes.Then,an improved coding parameter recognition scheme based on deep learning is proposed.Three network models,LSTM,Res Net,and Caps Net,are mainly discussed,and the model is improved based on the characteristics of the coding parameter identification problem.Taking the generation matrix recognition of 1/2 convolutional code as an example,the recognition performance of each model is verified through simulation.Finally,consider improving the performance of parameter recognition from the perspective of reducing the error rate of the initial received codeword.Based on three methods:wavelet transform,ensemble empirical mode decomposition,and singular value decomposition,time-frequency analysis of noisy modulation signals is performed to reduce noise power,thereby reducing system error rate and indirectly improving the performance of coding parameter recognition.The performance of the three noise reduction methods is compared from subjective and objective multi-angle simulations.The results show that the singular value decomposition method has the best performance.Through noise reduction,the accuracy of coding parameter recognition can be further improved.