Software Implementation of Fast List Decoder for PAC Codes

Программная реализация быстрого списочного декодера для PAC-кодов
Rui Lv, Hang Yin, Jingxin Dai, Yansong Lv, Yuhuan Wang
2024-06-20

SIMD parallelization (NEON, SSE, AVX256)fast list decoderfour-node fast list (FFL) decoding algorithmpolarization-adjusted convolutional (PAC) codessoftware high-throughput decoder
In this letter, a novel high throughput software polarization-adjusted convolutional (PAC) decoder based on the four-node fast list (FFL) decoding algorithm is proposed. To improve the parallel processing capabilities of the proposed decoder, we effectively parallel map the FFL decoding algorithm to processors by using the single-instruction-multiple-data (SIMD) instruction set to decode multiple frames of data in parallel. Moreover, the proposed decoder has sufficient generality to be implemented on X86 and ARM processors (NEON, SSE, and AVX256 instruction sets). Experimentations show that the proposed software PAC decoder can achieve 26.256 Mb/s.
1
A novel high-throughput software PAC decoder based on the four-node fast list (FFL) decoding algorithm is proposed.
2
Experimental results show the proposed software PAC decoder achieves a throughput of 26.256 Mb/s.
3
The decoder maps the FFL algorithm to processors using SIMD to decode multiple frames in parallel, improving parallel processing capabilities.
4
The implementation is general enough to run on x86 and ARM processors, supporting NEON, SSE, and AVX256 instruction sets.

Software implementation of a fast list decoder for polarization-adjusted convolutional (PAC) codes

High-throughput, parallel SIMD-mapped performance and portability of the four-node fast list (FFL) PAC decoding algorithm on x86 and ARM (NEON, SSE, AVX256) processors, including achieved throughput

Publication Details
Publication Date
2024-06-20
Journal
Publisher
ISSN
Cited by
2
Access Type
Author Information
Authors
Rui Lv
Hang Yin
Jingxin Dai
Yansong Lv
Yuhuan Wang
Explore further
Open the scid.ai AI chat with a ready-made request: it will find papers on a similar topic and help build a literature review.
Find similar papers in the chat
Make a presentation
100%