Towards systematic Luby transform codes: optimisation design over binary erasure channel
Seyed Masoud Mirrezaei
Abstract
Seyed Masoud Mirrezaei
Abstract
In this Letter, the authors introduce optimisation design technique of systematic Luby transform (SLT) codes over binary erasure channel (BEC). SLT codes are more efficient than Luby transform (LT) codes since the former ones can recover source symbols more quickly. There was some work about designing degree distribution in LT codes and it was shown that erasure probability barely impact performances of LT codes. Using degree distributions designed for LT codes directly in SLT codes does not provide satisfied performances. For the first time, they introduce erasure probability as a parameter to establish a linear programming model in SLT codes. By applying And–or Tree analysis, they propose the asymptotic analysis formulae of SLT codes over BEC. To achieve the goal of recovering source symbols as soon as possible, the authors' objective of optimisation is to minimise overhead. Simulations are offered to validate the performance of their optimisation result. Besides, their degree distribution outperforms other distributions with respect to bit error ratio.
OpenAlex reports 2 citations for this work. Citation counts describe recorded attention and do not establish research quality.
A contribution statement is not available in the OpenAlex record.
Method details are not available in the OpenAlex metadata.
Findings are not separately available in the OpenAlex metadata.
Limitations are not available in the OpenAlex metadata.
Application details are not available in the OpenAlex metadata.
In this Letter, the authors introduce optimisation design technique of systematic Luby transform (SLT) codes over binary erasure channel (BEC). SLT codes are more efficient than Luby transform (LT) codes since the former ones can recover source symbols more quickly. There was some work about designing degree distribution in LT codes and it was shown that erasure probability barely impact performances of LT codes. Using degree distributions designed for LT codes directly in SLT codes does not provide satisfied performances. For the first time, they introduce erasure probability as a parameter to establish a linear programming model in SLT codes. By applying And–or Tree analysis, they propose the asymptotic analysis formulae of SLT codes over BEC. To achieve the goal of recovering source symbols as soon as possible, the authors' objective of optimisation is to minimise overhead. Simulations are offered to validate the performance of their optimisation result. Besides, their degree distribution outperforms other distributions with respect to bit error ratio.
Key concepts: Luby transform code, Online codes, Binary erasure channel, Fountain code, Tornado code, Computer science, Erasure, Algorithm