使用效率内存部分并行译码器结构的QC - LDPC码

Low-Density Parity-Check (LDPC) codes are linear
block codes specified by very sparse parity-check matrix H
[1]. LDPC codes have attracted considerable attention due to
their near Shannon limit performance and inherently
parallelizable decoding scheme. Quasi-Cyclic LDPC (QCLDPC)
codes are well suited for hardware implementation
because of the regularity in their parity check matrices.
Recently, several classes of QC-LDPC codes [2-5] have been
proposed that can achieve comparable performance with
equivalent random LDPC codes. Among various LDPC
codes decoding algorithms, the Sum Product (SP) algorithm
has the best decoding performance. The modified Min-Sum
algorithm [6], which doesn't require any knowledge about
the channel parameters and offers comparable decoding
performance to SP algorithm, is preferred in low complexity
hardware implementation.
In general, LDPC codes achieve outstanding
performance only with large code word lengths (e.g.,
N≥ 2000 bits). Thus, the memory part normally dominates
the overall hardware of a LDPC codec. A memory efficient
serial decoder was presented in [7]. The decoding throughput
is less than 5.5Mbps per tile. Partially parallel decoder
architectures, which can achieve a good trade-off between
hardware complexity and decoding throughput, are preferred
in practice. In this paper, a memory efficient partially
parallel decoder architecture for high rate QC-LDPC codes is
proposed, which exploits the data redundancy of soft
messages in the Min-Sum decoding algorithm. Typically,
over 30% memory can be reduced.
In this paper, a rearranged Min-Sum LDPC decoding
procedure and the associated partially parallel decoder
architecture are proposed to reduce the required memory for
storing the extrinsic soft messages. To reduce the complexity
of Check-node Processing Unit (CPU), an optimized Pseudo
Rank Order Filter (PROF) is proposed. A low complexity
data scheduling structure is developed to enable parallel
processing. The required memory can be further reduced by
replacing the dual-port memory with single-port memory. In
this case, the simultaneous memory read and write
operations are performed at different memory segments
while employing memory partitioning and data arbitration
techniques [10].
The structure of this paper is as follows. In Section II,
The rearranged Min-Sum decoding procedure is discussed.
Section III presents the partially parallel decoder
architecture. Various optimizations to further reduce the
hardware complexity are addressed in Section IV. The
conclusions are drawn in Section V.