Dynamic Frozen-Function Design for Reed-Muller Codes With Automorphism-Based Decoding

In this letter, we propose to add dynamic frozen bits underlying polar codes with a Reed-Muller information set the aim of maintaining same sub-decoding structure in Automorphism Ensemble (AE) and lowering Maximum Likelihood (ML) bound by reducing number minimum weight codewords. We provide freezing constraint matrix that remains identical after applying permutation linear transformation. This feature also permits drastically reduce memory requirements an AE decoder polar-like having bits. show that, under decoding, proposed constraints lead gain up 0.25dB compared ML R(3,7) code, at cost small increase requirements.