Partitioned Adaptive Successive-Cancellation List Decoder for Polar Codes

This paper focuses on a generalization of the well-known adaptive successive cancellation list decoding decoding (ASCLD) of polar codes, characterized by several extra decoding attempts with larger list size, when all the previous decoding attempts fail. To further improve the adaptive efficiency, we consider use the partitioning encoding structure of polar codes concatenated with cyclic redundancy check (CRC) code. To improve CRC detecting efficiency and avoid the performance loss caused by undetecting, we design the partitioning structure based on classifying the decoding errors into two types. Compared the original ASCLD, the novel scheme can detect the decoding error more timely and reduce the decoding delay. Once decoding error is detected, decoders immediately decodes the current partition again with larger list size. Moreover, to further improve the performance, we design a new path restoring mechanism. From the decoding trajectories generated at the previous partition, all the ones that can pass CRC will be restored as the decoding input of the next partition. Simulation results show that the proposed scheme can reduce the delay of ASCLD with similar performance.