Fast SC-Flip Decoding of Polar Codes with Reinforcement Learning

In this paper, we introduce a novel bit-flipping algorithm for fast successive cancellation (FSC) decoding of polar codes. In particular, we first propose a new bit-flipping strategy tailored to single parity-check (SPC) constituent codes of polar codes. A parameterized bit-flipping model is then de...

Full description

Saved in:
Bibliographic Details
Main Authors: Doan, Nghia, Hashemi, Seyyed Ali, Ercan, Furkan, Gross, Warren J.
Format: Conference Proceeding
Language:English
Subjects:
Decoding
Error probability
fast SC flip
Heuristic algorithms
machine learning
polar codes
Prediction algorithms
Predictive models
Reinforcement learning
Supervised learning
Online Access:Request full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:In this paper, we introduce a novel bit-flipping algorithm for fast successive cancellation (FSC) decoding of polar codes. In particular, we first propose a new bit-flipping strategy tailored to single parity-check (SPC) constituent codes of polar codes. A parameterized bit-flipping model is then developed and reinforcement learning (RL) is used to optimize the parameters. Our experimental results show that for a polar code of length 512 with 256 information bits, the proposed decoder has a better or similar error-correction performance compared to the state-of-the-art fast DSCF (FDSCF) decoding algorithm when the same number of maximum decoding attempts is considered.
ISSN:1938-1883
DOI:10.1109/ICC42927.2021.9500904