一种极化码联合SC球形列表译码算法

为了降低极化码译码复杂度,提出了一种极化码联合SC球形列表(joint successive cancellation sphere list,JSCSL)译码算法.JSCSL译码算法结合了串行消除列表(successive cancellation list,SCL)译码和球形列表(list sphere decoding,LSD)译码相同的列表结构,联合不同的译码流程,增加译码并行性,从而降低译码复杂度.JSCSL译码算法将码字分割,在2个方向上使用不同译码方案,即从前往后和从后往前分别采取SCL译码和LSD译码,最终选取满足冗余循环校验(cyclic redundancy check,CRC)校验的码字.理论推导了2部分译码的中间分割点,使得独立译码时间近似,以达到最优的并行译码.仿真结果表明,与性能较优的CA-SCL(CRC-aided successive cancella-tion list)译码算法相比,JSCSL译码算法性能与其相近,但译码复杂度降低了40％～50％,因此,JSCSL译码算法在译码性能和复杂度达到新的平衡.

List polar decoder jointed with successive cancellation and sphere decoding

In order to reduce the decoding complexity of polar code, the joint successive cancellation sphere list (JSCSL) decoding algorithm is proposed in this paper. The JSCSL decoding algorithm combines the same list of successive cancellation list (SCL) decoding and list sphere decoding (LSD) decoding, combines different decoding processes to increase the decoding parallelism, and then reduces decoding complexity. The JSCSL decoding algorithm divides the codewords into two parts and uses different decoding schemes in opposite direction, i.e., SCL decoding is adopted from the front to the back and LSD is adopted from the back to the front respectively. Finally, the codewords that satisfies the cyclic redundancy check (CRC) is selected. The theoretical derivation of the intermediate segmentation points of the two-part decoding makes the independent decoding time approximate to achieve the optimal parallel decoding. The simulation results show that the performance of the JSCSL decoding algorithm is close to CRC-aid successive cancellation list (CA-SCL) decoding algorithm, but the decoding complexity is reduced by 40%~50%. Therefore, the JSCSL decoding algorithm achieves a new balance in decoding performance and complexity.