Efficient stochastic successive cancellation list decoder for polar codes

Polar codes are one of the most favorable capacity-achieving codes owing to their simple structures and low decoding complexity. Successive cancellation list (SCL) decoders with large list sizes achieve performances very close to those of maximum-likelihood (ML) decoders. However, hardware cost is a severe problem because an SCL decoder with list size L consists of L copies of a successive cancellation (SC) decoder. To address this issue, a stochastic SCL (SSCL) polar decoder is proposed. Although stochastic computing can achieve a good hardware reduction compared with the deterministic one, its straightforward application to an SCL decoder is not well-suited owing to the precision loss and severe latency. Therefore, a doubling probability approach and adaptive distributed sorting (DS) are introduced. A corresponding hardware architecture is also developed. Field programmable gate array (FPGA) results demonstrate that the proposed stochastic SCL polar decoder can achieve a good performance and complexity tradeoff.