A novel combinatorial framework to construct spatially-coupled codes: Minimum overlap partitioning


Esfahanizadeh H., HAREEDY A., Dolecek L.

2017 IEEE International Symposium on Information Theory, ISIT 2017, Aachen, Germany, 25 - 30 June 2017, pp.1693-1697 identifier identifier

  • Publication Type: Conference Paper / Full Text
  • Doi Number: 10.1109/isit.2017.8006818
  • City: Aachen
  • Country: Germany
  • Page Numbers: pp.1693-1697

Abstract

© 2017 IEEE.Spatially-coupled (SC) codes are a family of graph-based codes that have attracted significant attention thanks to their capacity approaching performance. An SC code is constructed by partitioning an underlying block code into a number of components, and coupling their copies together. The number of components is determined by the memory parameter. In this paper, we study a finite length construction for the circulant-based SC codes. We introduce a new partitioning scheme, which we call minimum overlap partitioning, that outperforms previous methods. We also present a general approach for the enumeration of problematic objects in the error-floor regime that can be applied to any circulant-based SC code and to a variety of partitioning schemes. Compared to the uncoupled block codes, an SC code constructed by the new approach has more than 1.5 and 3 orders of magnitude performance improvement for the memory 1 and 2, respectively. Additionally, it outperforms the existing method of partitioning via cutting vectors by at least half an order of magnitude; this performance advantage becomes more pronounced for SC codes with higher memories.