Analysis on crossover probability estimation using LDPC syndrome

作者：Fang Yong^*

来源：Science China Information Sciences, 2011, 54(9): 1895-1904.

DOI：10.1007/s11432-011-4311-y

• distributed source coding
• Slepian-Wolf coding
• LDPC syndrome
• correlation estimation

Correlation estimation is a critical issue that impacts the application of Slepian-Wolf coding (SWC) in practice. Dynamic online correlation estimation is a type of newly-appearing approaches, in which the decoder estimates the virtual correlation channel between two correlated sources using both side information and the compressed SWC bitstream of the source. Since the compressed SWC bitstream usually contains partial information of the source, the emergence of dynamic online correlation estimation is helpful to solving the problem of correlation estimation in the SWC and further makes the SWC realisable. Currently, the SWC is usually implemented by LDPC codes. In this case, the SWC bitstream is just the LDPC syndrome of the source. It has been revealed that there are residual redundancies in LDPC syndromes, which can be used to estimate the crossover probability between two correlated binary sequences. However, this algorithm has not been well justified yet. This paper makes use of the central limit theorem (CLT) to establish a mathematic model for analyzing the performance of this algorithm. Especially, for irregular LDPC codes, the optimization of weight vectors is discussed in detail. Representative experimental results are provided to validate the analysis.

• 出版日期2011-9
• 单位西北农林科技大学