Stepped List Decoding for Polar Codes

In the successive cancellation list (SCL) decoding of polar codes, as the list size increases, the error correction performance improves. However, a large list size results in high computational complexity and large memory requirement.In this paper, we investigate the list decoding process by introd...

Full description

Saved in:
Bibliographic Details
Main Authors: Rowshan, Mohammad, Viterbo, Emanuele
Format: Conference Proceeding
Language:English
Subjects:
Computational complexity
Decoding
Error correction
Iterative decoding
Magnetic resonance imaging
Measurement
Memory management
memory requirement
Polar codes
successive cancellation list decoding
Online Access:Request full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:In the successive cancellation list (SCL) decoding of polar codes, as the list size increases, the error correction performance improves. However, a large list size results in high computational complexity and large memory requirement.In this paper, we investigate the list decoding process by introducing a new parameter named path metric range (PMR) to elucidate the properties of the evolution of the path metrics (PMs) within the list throughout the decoding process. Then, we advocate that the list size can change stepwise depending on PMR. As a result, we propose a stepped list decoding scheme in which the error correction performance of the conventional list decoding is preserved while the path memory may reduce by 75%, the size of the internal LLR memory and partial sums memory can drop by 50%, and the computational complexity may halve. The reduction in complexity is SNR-independent and achieved without introducing any computational overhead.
ISSN:2165-4719
DOI:10.1109/ISTC.2018.8625267