Progressive Construction of Two-Dimensional Polar Codes With Path-Selective Decoder

Polar codes are extensively employed in Ultra-Reliable Low Latency (URLL) scenarios due to their near Maximum Likelihood (ML) performance. This letter introduces a novel two-dimensional polar coding scheme paired with a highly parallelized decoder. To construct the codes within this scheme, we present a refined frozen bits selection algorithm termed Double Gaussian Approximation (DGA), enabling precise estimation of Signal-to-Noise Ratio (SNR) conditions across subchannels. Additionally, we develop a Selective Successive Cancellation List (SSCL) decoder for efficient decoding of both inner and outer codes. Simulation results demonstrate that the proposed scheme surpasses polar codes utilized in the 5G standard by 0.1dB, accompanied by an approximately 20% reduction in latency and decoding complexity.