Numerical Issues Affecting LDPC Error Floors
Brian K. Butler, Paul H. Siegel
TL;DR
This paper investigates how numerical issues impact the error floors in LDPC decoders during floating-point simulations and proposes computational solutions to mitigate these issues, significantly reducing error floors.
Contribution
It introduces methods to handle numerical problems in belief propagation decoders, improving the accuracy of error floor estimation in LDPC codes.
Findings
Careful message processing reduces error floors by orders of magnitude.
Proposed solutions improve the sum-product algorithm's numerical stability.
Handling highly-certain messages is crucial for accurate LDPC decoding.
Abstract
Numerical issues related to the occurrence of error floors in floating-point simulations of belief propagation (BP) decoders are examined. Careful processing of messages corresponding to highly-certain bit values can sometimes reduce error floors by several orders of magnitude. Computational solutions for properly handling such messages are provided for the sum-product algorithm (SPA) and several variants.
Peer Reviews
No public reviews on file for this paper yet. If you reviewed it on a platform where reviews are public (OpenReview, ICLR, NeurIPS, ICML), you can paste yours below so the community can read it here.
Videos
No videos yet. Explain this paper in a talk, walkthrough, or lecture? Add one.
Taxonomy
TopicsError Correcting Code Techniques · Advanced Wireless Communication Techniques · Wireless Communication Security Techniques