Turbo polar code based on soft-cancelation algorithm

Abstract

Since the first polar code of Arikan, the research field of polar codes has been continuously active. Improving the performance of finite-code-length polar codes is the central point of this field. In this paper, the parallel concatenated systematic turbo polar code (PCSTPC) model has been proposed to improve the polar codes performance in a finite-length regime. On the encoder side, two systematic polar encoders are used as constituent encoders. While on the decoder side, two single iteration soft-cancelation (SCAN) decoders are used as soft-in-soft-out (SISO) algorithms inside the iterative decoding algorithm of the parallel concatenated systematic turbo polar code (PCSTPC). As compared to the optimized turbo polar code with SCAN and BP decoders, the proposed model has about 0.2 dB and 0.48 dB gains at BER=10^(-4), respectively, in addition to 0.1 dB, 0.31 dB, and 0.72 dB gains over the TPC-SSCL32, TPC-SSCL16, and TPC-SSCL8 models, respectively. Moreover, the proposed model offers less complexity in comparison with other models, therefore requiring less memory and time resources.