Abstract:

The existing parameter recognition algorithms for non-punctured polar codes default the frozen bit values to 0, and do not have the ability to judge the frozen bit values. Aiming at this problem, the parity check relation between codeword matrix and column vectors of generator matrix is deduced, where the identification of bit type and the judgment of frozen bit value are converted into a test of the check relationship. Based on the above, the average likelihood difference is introduced as a statistic volume, and the theoretical distribution and decision threshold of the statistic volume are given. The algorithm uses the proportion of information positions in the decision result to identify the parameters such as code rate and code length. Simulation experiments show that the derivation of the relevant properties is consistent with the actual situation. The proposed algorithm overcomes the shortcomings of existing methods that cannot identify frozen bit values on the basis of the same computational complexity level and without reducing the recognition performance. When the signal to noise ratio is 5.5 dB, the recognition accuracy is still close to 100% for polar codes with a code length of 512 of the polar code.

Key words: non-cooperative communication, polar code, non-punctured, recognition