Simplified successive-cancellation decoding using information set reselection for polar codes with arbitrary blocklength

The distribution of information bits and frozen bits on the decoder graph can be exploited to simplify the successive-cancellation (SC) decoding of polar codes. In this study, the authors establish a general simplified SC decoding framework for polar codes with arbitrary block-length, which is independent of any specific or explicit generator matrices and considerably reduces decoding complexity while retaining the same error performance. Then, based on the fact that the complexity reduction depends on the distribution of information bits, a so-called m-radius reselection scheme is proposed to construct multiple feasible information sets which are different from the original one defined by Ari{dotless}kan. In this way, flexible tradeoffs between performance and complexity could be achieved.