Recursive Trellis Processing of Large Polarization Kernels

A reduced complexity algorithm is presented for computing the log-likelihood ratios arising in the successive cancellation decoder for polar codes with large kernels of arbitrary dimension. The proposed algorithm exploits recursive trellis representation of the codes generated by submatrices of the polarization kernel, and enables codes based on large kernels to provide better performance compared to the codes based on Arikan kernel with the same decoding complexity.
Comment: Submitted to ISIT2021