Hybrid Decoding of CRC-Polar Codes

Abstract

Successive cancellation list (SCL) decoding of polar codes can yield excellent performance in error correction with the assistance of cyclic redundancy check (CRC) codes while its complexity remains high. This paper proposes a hybrid decoding for the CRC assisted polar codes, namely the CRC-polar codes. The proposed mechanism integrates the Fano decoding and the successive cancellation (SC) decoding. The Fano decoding will be first deployed to recover the message. However, when the received information is unreliable, it may stagger causing a large latency. To overcome this, we introduce a Fano decoding computation threshold. If a CRC validated estimation cannot be provided within the threshold, the SC decoding will be further deployed. As a result, both the decoding complexity and latency can be contained. The threshold-performance tradeoff insight of the hybrid decoding is also studied. Our simulation results demonstrate that the proposed hybrid decoding outperforms the SCL and the Fano decoding with a lower complexity.