Abstract:

In order to solve the problem that the current polarization code parameter recognition algorithm lacks the recognition for the starting point of code word and the information bit recognition algorithm is complicated, a polarization code parameter blind recognition algorithm based on the code matrix estimation is proposed. The proposed algorithm firstly multiplies the intercepted codeword matrix with the Kronecker matrix under the corresponding code length and the reverse rearrangement matrix to obtain the code matrix estimation, then identifies the code length and the code word starting point by the distribution characteristics of the code matrix, and finally uses the trained convolutional neural network to identify the information bits and freezing bits of the polarization code. The experimental results show that the proposed method not only realizes the recognition for the starting point of code word, but also realizes the code length recognition when the code word's starting point is unknown, and the code length recognition accuracy is better than the existing algorithm. When the bit error rate is 0.19, the polarization code length recognition rate of parameter (32, 12) can still reach more than 90%.

Key words: polarization code, parameter blind identification, encoding estimation matrix, neural network