IEEE TRANSACTIONS ON MAGNETICS, VOL. 50, NO. 11, NOVEMBER 2014 3101304
Improved Min-Sum Decoding for 2-D
Intersymbol Interference Channels
Lingjun Kong
1,2
, Yunxiang Jiang
2
, Guojun Han
3
,FrancisC.M.Lau
2
, and Yong Liang Guan
4
1
School of Computer Science and Technology, Nanjing University of Posts and Communications, Nanjing 210003, China
2
Department of Electronic and Information Engineering, Hong Kong Polytechnic University, Hong Kong
3
School of Information Engineering, Guangdong University of Technology, Guangzhou 510006, China
4
School of Electrical and Electronic Engineering, Nanyang Technological University, Singapore 639798
In this paper, 2-D normalized min-sum (NMS) algorithm and offset min-sum (OMS) algorithm are proposed for efficient decoding
of low-density parity-check (LDPC) codes in 2-D intersymbol interference (ISI) ultra-high density magnetic recording channels,
such as bit-patterned magnetic recording and 2-D magnetic recording, where a reduced-complexity 2-D detector based on the
iterative row-column soft detection feedback with Gaussian approximation detector is employed instead of the full 2-D Bahl–Cocke–
Jelinek–Raviv (BCJR) detector. The normalization and offset factors of the 2-D NMS and 2-D-OMS, are optimized based on the
extended density evolution for LDPC coded 2-D ISI channel, respectively. Simulation results show the performance loss caused by
the reduced-complexity LDPC decoder can be almost fully recovered by the proposed approaches, while retaining the benefit of low
complexity in decoder compared with the belief propagation (BP) decoding. Furthermore, both the NMS and OMS exhibit a lower
error floor than that of BP decoding in high signal-to-noise ratio region.
Index Terms—2-D intersymbol interference (ISI) channels, density evolution (DE), low-density parity-check (LDPC) codes, min-sum
algorithm (MSA).
I. INTRODUCTION
D
UE TO continued strong demand for increasing stor-
age areal densities, bit-patterned media recording, heat
(or microwave) assisted magnetic recording, and shingled
writing/2-D magnetic recording have been proposed to push
the storage density of future hard disks beyond 1 Tb/in
2
[1].
However, at such ultra-high recording density, many new
challenges from the magnetic recording medium, read/write
head as well as writing and reading channels arise. From a
signal processing and coding point of view, the most important
technical challenge which severely degrades the performance
of data detection is the 2-D intersymbol interference (ISI),
which is constituted by both 1-D ISI and intertrack interference
caused by the reduction of the track pitch [2], [3].
Although the symbol-based BCJR 2-D channel detector
[4], [5]—hereinafter called full 2-D channel detector
(equalizer)—can provide optimal detection over such 2-D ISI
channels, it has prohibitive complexity for the typical 2-D data
size. Suboptimal 2-D channel detectors with lower complexity
have been proposed in [5]–[8], but their computational burden
remains a bottleneck of the receiver. In [9], the well-known
suboptimal 2-D channel detector, iterative row-column soft
decision feedback algorithm (IRCSDFA) [7], has been further
simplified using Gaussian approximation (GA), the resultant
simplified 2-D detector is referred as IRCSDFA-GA. It suf-
fers performance loss compared with the original IRCSDFA
(as expected), but the loss is minimized by careful decoding-
detection scheduling design.
In this paper, we focus on decoding in the 2-D ISI channels
to compensate for performance loss that is inflicted using
Manuscript received March 2, 2014; accepted April 9, 2014. Date of
current version November 18, 2014. Corresponding author: L. Kong (e-mail:
ljkong@njupt.edu.cn).
Color versions of one or more of the figures in this paper are available
online at http://ieeexplore.ieee.org.
Digital Object Identifier 10.1109/TMAG.2014.2317749
the reduced-complexity low-density parity-check (LDPC)
detector, and present the 2-D normalized min-sum (NMS) and
offset min-sum (OMS) algorithms in 2-D ISI channel using
the IRCSDFA-GA detector. Specifically, the normalization and
offset factors of 2-D NMS and 2-D OMS, respectively, are
optimized based on the extended density evolution (DE) for
2-D ISI channels. By analyzing noise threshold with DE,
a parallel differential optimization (PDO) presented in [10]
is considered to obtain the normalization and offset factors.
In so doing, we show that the performance loss incurred by
the reduced-complexity min-sum (MS) decoder can be almost
fully recovered by the proposed decoding algorithms compared
with the belief propagation (BP) decoding. An additional
useful feature of the proposed approaches is that it has a lower
error floor relative to the original BP decoding in 2-D ISI
channels.
The rest of this paper is organized as follows. Section II
introduces the 2-D ISI channel model and reduced-complexity
IRCSDFA-GA detector. Then, in Section III, 2-D NMS and
2-D OMS for LDPC coded 2-D ISI channels are proposed
based on the extended DE using the IRCSDFA-GA detector.
The proposed schemes are validated in Section IV. Finally,
Section V presents the conclusion.
II. 2-D ISI C
HANNELS AND DETECTOR
In this section, we introduce the 2-D ISI channel model and
briefly review the technique of IRCSDFA-GA detection over
GA-simplified ISI trellis reported in [9].
A. 2-D ISI Channels Model
Let x
(i, j)
∈{−1, 1} denote the binary data distributed in an
array with M rows and N columns, with i = 1, 2,...,M,
and j = 1, 2,...,N. L
M
and L
N
denote the interference
lengths in the horizontal and vertical directions, respectively.
Furthermore, let the 2-D channel response before equalization
0018-9464 © 2014 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.