Asymmetric Slepian-Wolf Coding of Nonstationarily-Correlated M-ary Sources with Sliding-Window Belief Propagation

Abstract

Asymmetric Slepian-Wolf coding (ASWC) of M -ary sources with nonstationary correlation is a very useful model for many practical problems. An effective implementation of this coding scheme is to binarize each M -ary source into multiple bitplanes which are then compressed by a single binary lowdensity parity-check (LDPC) code. Though the inter-bitplane correlation of M -ary sources can be exploited at the decoder by the joint-bitplane belief propagation (JBBP) algorithm, accurate online estimation of varying local source correlation still remains a major challenge. To tackle this problem, this paper proposes an M -ary counterpart of the sliding-window belief propagation (SWBP) algorithm to realize simultaneous source recovery and correlation estimation. To search for the optimal sliding-window size, the expected rate is raised as a new criterion. Moreover, an adaptive method is proposed to decide whether correlation reestimation is necessary. The M -ary SWBP (MSWBP) algorithm is then generalized to obtain its 2D form, which can be used to tackle the ASWC of 2D M -ary sources with nonstationary correlation. The developed 1D/2D-MSWBP algorithm inherits all merits of the original binary SWBP algorithm, e.g., near-optimal coding efficiency, low complexity, insensitivity to initial settings, etc., making it a very attractive technique in practice.

Graphics processing unit-accelerated joint-bitplane belief propagation algorithm in DSC

“…Suppose {Y i = y i }, the residual Z i can be got from [−y i : (M − y i )) so that Z i is dependent on Y i . As long as this virtual channel can be modeled by some standard communication channel, e.g., the truncated discrete Laplace (TDL) distribution and the truncated discrete Gaussian (TDG) distribution [15], channel codes can be effectively employed. Since those two distributions have no affect on the acceleration of JBBP on GPU, the model TDL distribution is used in this paper.…”

“…As shown in Fig. 3 [15], the factor graph of the JBBP algorithm includes three kinds of nodes: SNs (marked by ellipses), BNs (marked by circles), and CNs (marked by squares). The decoder should be initialized through two stages before running the JBBP algorithm.…”

“…The decoder should be initialized through two stages before running the JBBP algorithm. In addition, the original JBBP algorithm is proposed in the literature [40] and the computational complexity is reduced while remaining the same coding efficiency in the literature [15]. So our work is based on the simplified algorithm.…”

“…An adaptive DVC framework with expectation propagation correlation estimator is proposed and it is carried out in pixel-domain using JBBP algorithm for decoding [9]. The JBBP algorithm is utilized at the decoder to exploit the inter-bitplane correlation of M-ary sources, and its computational complexity is reduced to some extent in [15]. The JBBP algorithm has an abroad application as a useful decoding method of DSC, but it suffers from intensive computation for running the JBBP algorithm whose complexity is significantly higher than BP algorithm and not only that, DSC enables low-complexity encoding where the bulk of the computation is shifted to the decoder so that the decoder is faced with the problem of high complexity.…”

Hamming Distance Spectrum of DAC Codes for Equiprobable Binary Sources

“…Recently, AC codes also find their application to lossless distributed source coding (DSC), or Slepian-Wolf coding (SWC) [10], which has traditionally been implemented with channel codes, e.g., turbo codes [11] and low-density paritycheck (LDPC) codes [12], [13], [14]. Such solutions are known as distributed AC (DAC) codes.…”