Enhanced Successive Cancellation List Decoding of Polar Codes

Abstract

Polar codes can be efficiently decoded by the successive cancellation list (SCL) algorithm in which the sorter plays an important role. The sorter, however, is time-consuming. We significantly improve this decoding scheme by developing a new non-sorting direct selection method for the SCL algorithm. In the non-sorting strategy, we design a new selector for choosing the $L$ best surviving paths, which is achieved by exploiting the bitwise comparison of the candidate path metrics. The analysis and simulation results show that, for $M$ -bit length path metrics, our direct selection method with time complexity $O(ML)$ ) is superior to the latest simplified bubble sorter and the quick sorting strategy whose time complexity is $O(ML^{2})$ and $O(ML \log L)$ , respectively. In the scenario of short codeword length and large list size, our selector can remarkably reduce the latency and the hardware resource consumption while not assuming that the metrics in the decoding are sorted.