Skip to main navigation Skip to search Skip to main content

On the Minimum Distance and Erasure Correction of Codes with Block Circulant Topology

Research output: Chapter in Book/Report/Conference proceedingConference PaperResearchpeer-review

Abstract

Codes with locality without any global paritycheck constraints apart from those generated by local codes' constraints have recently found a unique application in decentralized systems. In response to this, we proposed in previous work, a new class of block circulant (BC) codes possessing certain structure in the arrangement of parity-check constraints referred to as a block circulant topology. The BC topology T[μ, λ, ω] (ρ) and BC codes CBC[μ, λ, ω, ρ] are parameterized by integers λ ≥ 2, ω ≥ 2, ρ ≥ 2 and μ a multiple of λ. In this work, we show that the rate and the minimum distance of the BC code scale with corresponding metrics of its local codes in a manner that can not be realized by well-known linear array and product topologies, thus widening the possible regime of operation. We also provide an efficient erasure-correcting decoder for CBC[μ, λ=3, ω, ρ], while such a decoder was earlier known only for λ=2. The decoding algorithm uses a novel mechanism that iteratively corrects erasures from either a single or a triplet of local codes. We show that the minimum distance of CBC[μ, λ=3, ω, ρ] is 3 ρ+1, whereas the same result was earlier available under a constraint that μ=2a · 3 for some integer a.

Original languageEnglish
Title of host publicationISIT 2025 - 2025 IEEE International Symposium on Information Theory, Proceedings
EditorsAchilleas Anastasopoulos, Vincent Tan, Michele Wigger, Mark Wilde
Place of PublicationPiscataway NJ USA
PublisherIEEE, Institute of Electrical and Electronics Engineers
Number of pages6
ISBN (Electronic)9798331543990
ISBN (Print)9798331543990
DOIs
Publication statusPublished - 2025
EventIEEE International Symposium on Information Theory 2025 - Ann Arbor, United States of America
Duration: 22 Jun 202527 Jun 2025

Publication series

NameIEEE International Symposium on Information Theory - Proceedings
PublisherIEEE, Institute of Electrical and Electronics Engineers
ISSN (Print)2157-8095
ISSN (Electronic)2157-8117

Conference

ConferenceIEEE International Symposium on Information Theory 2025
Abbreviated titleISIT 2025
Country/TerritoryUnited States of America
CityAnn Arbor
Period22/06/2527/06/25

Cite this