An Improved Critical Set for List Decoding of Polar Codes
Heng Liu, Jin Sha, Xiumin Wang
Abstract
Polar codes have received significant attention as part of the 5G standard. The successive cancellation list (SCL) decoding can provide superior decoding performance compared with successive cancellation (SC) decoding. To improve the performance of SCL decoding, the key is to obtain the indices of the bit-channels which contribute the most to the elimination of the correct path. In this letter, we analyze the distribution of bit positions where the correct path is eliminated and propose an offline improved critical set (ICS) that obtains the index of the erroneous bit for SCL decoding. Numerical results show that ICS is an effective alternative to the conventional critical set (CS) for SCL decoding and the average computational complexity is reduced significantly without performance loss compared with CS.