On Multibasis Information Set Decoding
Sebastian Bitzer, Martin Bossert · 2022 IEEE International Symposium on Information Theory (ISIT) · 2022
Information set decoding is a method for soft-decision decoding of general linear binary codes. Its performance can be improved by reprocessing multiple bases. Different methods for choosing the bases are known. We present a novel method for basis selection using probability analysis. The sequence of bases is determined which maximizes the decoding performance. We present a method for approximating this sequence by updating the error probabilities of the received symbols and give an efficient implementation. Furthermore, we show that the concept of updating bit error probabilities can be extended from information set decoding to box and match decoding. Simulation results confirm the efficiency of the proposed decoders compared with regular information set decoding and other multibasis algorithms.