You are on page 1of 7

120 IEEE TRANSACTIONS ON BROADCASTING, VOL. 54, NO.

1, MARCH 2008

DVB-S2 LDPC Decoding Using Robust Check Node Update Approximations


Stylianos Papaharalabos, Marco Papaleo, Student Member, IEEE, P. Takis Mathiopoulos, Senior Member, IEEE,
Massimo Neri, Member, IEEE, Alessandro Vanelli-Coralli, Senior Member, IEEE, and
Giovanni E. Corazza, Member, IEEE

Abstract—Broadband satellite services to fixed terminals are transmission signals, such as the Amplitude Phase Shift Keying
currently offered in the forward link by the 2nd Generation (2G) (APSK) signaling. It is interesting to note that the resulting
Digital Video Broadcasting Satellite (DVB-S2) standard. For this capacity increase for the new DVB-S2 over DVB-S standard
standard the use of powerful Low-Density Parity-Check (LDPC) has been approximately 30% in broadcasting mode and more
error correcting codes has been adopted performing within ap-
than 100% for unicasting (i.e. broadband) interconnections
proximately 1 dB from the Shannon capacity limit. This paper
studies and compares for the first time in a systematic manner [9], [10]. Nowadays, the provision of digital broadcasting
different approximation methods used in check node update multimedia contents to mobile users has been a new emerging
computation of DVB-S2 LDPC decoding with the aim of reducing technology. For example, a special journal issue entitled ‘Mo-
computational complexity. Various performance evaluation re- bile Multimedia Broadcasting’ focusing mainly on terrestrial
sults are presented for a wide range of DVB-S2 parameters, such systems has been recently published [11]. In this journal issue,
as LDPC codeword size, coding rate, modulation format and in- it is noted that circulant LDPC codes have been adopted to
cluding several decoding algorithms. It is shown that the proposed provide Digital Terrestial/Television Multimedia Broadcasting
check node update approximations have a robust behavior, i.e. (DTMB) services in China [12], [13]. The evolution of the new
the resulting performance is quite independent of the DVB-S2
modulation and coding parameters. It is further shown that these DVB-S2 standard to support fully mobility is another research
approximations perform very close to the optimal Sum-Product topic currently under investigation [14]. It aims at offering
Algorithm (SPA) in degradation, which is less than 0.2 dB. Despite broadband interactive services to mobile users situated on
this small degradation, the reduction in computational complexity aircrafts, ships and high-speed trains targeting a niche market.
compared to the optimal SPA is significant and can be as high as LDPC codes can be decoded by the Sum-Product Algorithm
40% in computational time savings. (SPA), which is based on the message passing principle [3]. To
Index Terms—Broadcasting, iterative decoding, Low-Density reduce its complexity several decoding algorithms have been
Parity-Check (LDPC) codes. proposed over the past years. A detailed review of the currently
available SPA reduced complexity decoding algorithms for
LDPC codes can be found in [15]. Two practical implementa-
I. INTRODUCTION tion aspects of DVB-S2 LDPC decoding were proposed in [16]
reducing the number of iterations as well as the computation

L OW-DENSITY parity-check (LDPC) codes were pro-


posed initially by Gallager in the early 1960s [1] but they
were not used for many years mainly because the technology
operations. Some recent hardware implementations of the
DVB-S2 LDPC codec can be found in [17]–[20]. Among the
critical technical issues addressed in these references were the
was not mature for their practical implementation. In 1981 encoder/decoder architecture, the area size and the achieved
Tanner proposed a graphical representation of these codes, overall throughput. A DVB-S2 digital receiver was reported
also known as Tanner graphs [2]. However, it was not until in [21] dealing with several practical aspects, such as frame
the late 1990s when they were rediscovered by MacKay [3] syncronization, timing acquisition, signal-to-noise ratio (SNR)
and others [4]–[6]. LDPC codes can achieve near Shannon estimation and digital predistortion. It should be noted that
capacity limit performance over the Binary Erasure Channel although in the previously mentioned references there have
(BEC) as well as over the Additive White Gaussian Noise been various performance evaluation results published, this
(AWGN) channel with moderate decoding complexity [3], [7]. was done in a rather ad-hoc manner without using a systematic
Recently, they have been adopted by the new version of the approach.
Digital Video Broadcasting by Satellite (DVB-S2) standard Motivated by this observation, the purpose of this paper is
[8]. In this standard, an inner LDPC code is incorporated to present for the first time a systematic and thorough perfor-
with Adaptive Coding and Modulation (ACM) techniques as mance evaluation study of DVB-S2 LDPC codes with different
well as optimized high order modulation schemes for satellite modulation and coding parameters in conjunction with reduced
complexity decoding algorithms. Following [22], we have used
either a Piecewise (PW) linear function with a few entries or
Manuscript received August 1, 2007; revised October 15, 2007. This work a small Look-Up Table (LUT) to approximate the check node
was supported in part by IST SatNEx-II FP6 Project (IST-027393). update computation as accurate as possible and applied them to
S. Papaharalabos and P. T. Mathiopoulos are with the Institute for Space DVB-S2 LDPC decoding. It is shown that the adopted approxi-
Applications and Remote Sensing (ISARS), National Observatory of Athens,
15236 Athens, Greece (e-mail: spapaha@space.noa.gr; mathio@space.noa.gr). mations, which were optimized for pseudo-random based LDPC
M. Papaleo, M. Neri, A. Vanelli-Coralli and G. E. Corazza are with the De- codes in [22], offer two potential advantages when applied to
partment of Electronics, Computer Science and Systems (DEIS), University of DVB-S2 systems: (i) robust behavior that is independent on the
Bologna, 40136 Bologna, Italy (e-mail: mpapaleo@deis.unibo.it; mneri@deis. modulation and coding parameters, and (ii) good trade-off be-
unibo.it; avanelli@deis.unibo.it; gecorazza@deis.unibo.it).
Color versions of one or more of the figures in this paper are available online tween performance and complexity.
at http://ieeexplore.ieee.org. The structure of this paper is organized as follows. After this
Digital Object Identifier 10.1109/TBC.2007.911365 introduction, a brief overview of coding and modulation used in
0018-9316/$25.00 © 2007 IEEE
IEEE TRANSACTIONS ON BROADCASTING, VOL. 54, NO. 1, MARCH 2008 121

DVB-S2 standard is presented in Section II. Then, various algo- the check nodes corresponding to the parity-checks of a bipartite
rithms (i.e. both optimal and sub-optimal) used to decode LDPC graph and is the bit nodes corresponding to the encoded bits.
codes are described in Section III. The adopted approximations Moreover, each bit node is connected to check nodes and each
used in check node update computation of DVB-S2 LDPC de- check node is connected to bit nodes, i.e. a regular LDPC
coding are introduced in Section IV. Various performance eval- code is being considered. In following, denotes the set of
uation results assuming the DVB-S2 LDPC code with different check nodes connected to bit node and denotes the set
modulation and coding parameters are presented in Section V of bit nodes participating in the -th parity-check equation. In
and finally conclusions are made in Section VI. addition, represents the set , excluding the -th
check node and represents the set , excluding
the -th bit node.
II. CODING AND MODULATION FOR THE DVB-S2 STANDARD Consider now that a codeword of size is Binary Phase
Exploiting the increased feasibility of Application Specific Shift Keying (BPSK) modulated and transmitted over the
Integrated Circuit (ASIC) technology, the DVB-S2 committee AWGN channel. Also, is the received codeword corre-
has chosen to adopt a solution where Forward Error Correction sponding to the transmitted codeword with the same size . In
(FEC) encoding is performed into three steps using: (i) an outer general, the sum-product algorithm is composed of three steps:
systematic Bose-Chaudhuri-Hocquenghem (BCH) code; (ii) an (i) initialization; (ii) iterative process and (iii) hard decision
inner LDPC code and (iii) a block bit-interleaver [8]. With refer- [15]. Step (ii) includes the check node and bit node updates,
ence to the outer encoding, BCH codes are well-known -error which are the core of the algorithm. These updates can be
correcting block codes that prevent the propagation of unde- computed from
tectable errors generated by the inner LDPC decoder.
LDPC codes are linear block codes with sparse parity-check
matrix characterized by columns and rows, where
is the information block length, is the codeword length and
is the redundancy length added by the encoder. The (1)
LDPC codes standardized by DVB-S2 are irregular with vari- and
able bit node degrees and produce codewords of either 64800 or
16200 bits, i.e. long and short size codewords, respectively. The
adoption of irregular codes is because bit nodes with high de- (2)
grees collect more information from their adjacent check nodes
so that they can be corrected after a small number of iterations.
DVB-S2 LDPC codes have a well-defined structure, which is where is the log-likelihood ratio (LLR) of the
imposed of an appropriate parity-check matrix construction, message that the -th check node sends to the -th bit node,
by producing systematic codes in a very simple encoding opera- based on all the bits checked by except , and is
tion, that is the use of Irregular Repeat Accumulate (IRA) codes the LLR of the message that bit node sends to check node ,
[23]. based on all the checks involving except , respectively. In
Even though parity-check matrices of LDPC codes are addition, is the a posteriori probability of each bit node
sparse, in general generator matrices needed for encoding after transmission through the channel, i.e.
may be not. The latter matrices can be derived using the classical where is the noise variance.In order to obtain a wide range
Gaussian elimination method. Furthermore, a non-sparse matrix of spectral efficiencies in DVB-S2 standard, 11 different coding
can generate both storage and encoding complexity problems. rates (1/4, 1/3, 2/5, 1/2, 3/5, 2/3, 3/4, 4/5, 5/6, 8/9, 9/10) are
For that reason, DVB-S2 has chosen to adopt methods that par- available together with four different modulation schemes.
tially solve this problem by restricting a sub-matrix of the parity- Apart from the well-known -ary Phase Shift Keying (PSK)
check matrix to be lower triangular and thus eliminating the modulation formats, such as Quadrature PSK (QPSK) and
need to derive a generator matrix [23]. In this way, linear en- 8-PSK, the hybrid amplitude and phase modulation format, re-
coding complexity is feasible. In more detail, the parity-check ferred to as -ary APSK, is mainly considered for two reasons:
matrix has the form of where is a staircase (i) increased bandwidth efficiency which is achieved for high
matrix with column weight equal to 2, and values of , such as and , and (ii) its ability
is a random sparse matrix of size with column to better cope with typical non-linear distortion introduced by
weight ranging between 3 and 13 depending on the coding rate. satellite transponders. Following the notation
Another important feature is the periodicity of the sub-matrix where denotes the number of constellation points on the
, in order to reduce memory storage requirements by a factor -th ring, the considered in the DVB-S2 standard 16-APSK
of . In particular, for a group of bit nodes the check nodes and 32-APSK modulation formats are composed of a double
connected to the first bit node need only to be specified, whereas ring (4, 12) and a triple ring (4, 12, 16), respectively. Note that
the check nodes connected to -th bit node with can be the constellation parameters are optimized as a function of the
determined by a pre-calculated formula [23]. In DVB-S2 the pa- LDPC coding rate [8]. In the case of 8-PSK, 16-APSK and
rameter is chosen. 32-APSK modulation formats, a simple block bit-interleaver is
added after the LDPC encoding, in order to increase the code
diversity. Coded bits are serially written column-wise into the
III. OPTIMAL AND SUB-OPTIMAL REPRESENTATIONS OF THE
interleaver, and then are serially read row-wise.
SUM PRODUCT ALGORITHM (SPA)
The above algorithm is also known as the rule [15].
Let be a binary LDPC code described by a sparse Apart from it, [15] provides a description of two more optimal
parity-check matrix of size , where is algorithms with different check node update computation than
122 IEEE TRANSACTIONS ON BROADCASTING, VOL. 54, NO. 1, MARCH 2008

TABLE I
PIECEWISE LINEAR APPROXIMATION OF THE tanh(1) FUNCTION [22]

TABLE II
LUT APPROXIMATION OF THE tanh(1) FUNCTION [22] Fig. 1. Comparison of the two approximations of the tanh(1) function based
on PW linear and LUT with the exact curve.

in (1). These are: (i) Gallager’s approach and (ii) Jacobian ap-
proach. In the first case, the check node update computation can
Fig. 2. Normalized computational time against the number of decoding iter-
be expressed as ations using different decoding algorithms for the DVB-S2 LDPC code. Short
N R
codeword of = 16200 bits, coding rate ( ) 2/3, 8-PSK modulation, AWGN
channel and a maximum of 70 decoding iterations.

where denotes the logical Exclusive OR (XOR) operator. The


logarithm in the right part of (4) can be stored in memory by
using a Two-Dimensional (2D) LUT, usually with eight entities,
(3) in which both the sum and difference of the two LLR values
and are required.
For the min-sum algorithm, the correcting factor from (4) is
where denotes the signum function, the use of summa- omitted, so that the check node update can be computed from
tions is preferred instead of multiplications and also another
function is used, i.e.
. In the second case, the check
node update computation is applied into pairs of incoming mes-
sages in recursive form. It is based on the fact that for two
random variables and , with LLR values and
respectively, the following formula exists [15] (5)

It can be observed that the LLR values from (5), i.e. for min-sum
algorithm, are always greater than those from (3), i.e. for SPA.
In the open technical literature there exist several approaches
that improve the performance of the min-sum algorithm at the
expense of some complexity increase. For instance, in [15]
two algorithms were proposed, namely normalized min-sum
(4) (NMS) and offset min-sum (OMS) algorithms. Both are based
IEEE TRANSACTIONS ON BROADCASTING, VOL. 54, NO. 1, MARCH 2008 123

Fig. 3. PER performance of DVB-S2 LDPC code, N = 16200 bits, over the AWGN channel, a maximum of 70 decoding iterations, different coding rates and
modulation formats. Four decoding algorithms are compared with each other: (i) PW linear approximation; (ii) LUT approximation; (iii) EX method (i.e. ideal
performance) and (iv) min-sum algorithm.

on finding the minimum as shown in (5), but the former makes (or equivalently the LUT size) was done in order to have a sim-
use of a constant normalization factor : ilar degree of discrimination to the LUT used in Log-MAP de-
coding of turbo codes, which makes use of eight stored values
[24].
In order to compare the two approximations, i.e. PW linear
and LUT, in terms of computational time savings, various com-
puter simulation experiments were run assuming the DVB-S2
LDPC code with different codeword sizes, coding rates and
(6) modulation formats. In this respect, Fig. 2 illustrates the nor-
malized computational time obtained for the following scenario:
whereas the latter makes use of a subtraction with a positive short DVB-S2 LDPC codeword of 16200 bits, coding rate
constant value : equal to 2/3, 8-PSK modulation, AWGN channel, and a max-
imum of 70 decoding iterations. For comparison purposes, the
computational time provided by the optimal Jacobian approach
and the sub-optimal min-sum algorithm are also depicted. The
curves shown in Fig. 2 have been normalized to the reference
case, that is the EX method with a maximum of 70 decoding
(7) iterations. It can be seen that with respect to the EX method,
the PW linear approximation is about 26% faster, whereas the
LUT approximation is about 40% faster, respectively. In addi-
tion, the computational time of the Jacobian approach is about
IV. CHECK NODE UPDATE APPROXIMATIONS FOR DVB-S2 12% lower as compared to LUT approximation, whereas the
LDPC DECODING min-sum is the fastest algorithm, e.g. about 20% faster than the
For the case of pseudo-random based LDPC codes, the ef- Jacobian approach. However, as it will be shown in the next Sec-
fect on the BER performance when approximating the check tion, there is a small BER performance degradation associated
node update computation from (1) with different methods was with the min-sum algorithm. As expected, the simulation time
investigated in [22]. In particular, two different approaches were increases linearly with the number of decoding iterations. Based
considered to compute the function and its inverse: (i) on our experiments it is important to note that the computational
PW linear function with seven regions, and (ii) LUT with eight time savings are not influenced significantly by the selection of
values. These two approximations of the function (e.g. LDPC codeword size, coding rate and modulation format.
see Tables I and II) together with the exact expression (EX) are
illustrated in Fig. 1. The adopted approximations provide com- V. PERFORMANCE EVALUATION RESULTS
putational complexity savings with respect to the corresponding In order to assess the performance of the two SPA approxima-
EX expressions. The PW linear approximation requires only tions, the LDPC code considered in the DVB-S2 standard was
one multiplication and one addition, whereas the LUT approx- adopted. Thus, following [8] an outer BCH code is assumed,
imation requires no operations at all, apart from extra memory which can detect and successfully correct up to 12 erroneous
storage of eight values. The selection of the number of regions bits. The user packet length is 1504 bits (i.e. MPEG format),
124 IEEE TRANSACTIONS ON BROADCASTING, VOL. 54, NO. 1, MARCH 2008

Fig. 4. BER performance of DVB-S2 LDPC code, N = 16200 bits. Other system parameters are the same as in Fig. 3.

Fig. 5. PER performance of DVB-S2 LDPC code, N = 64800 bits. Other system parameters are the same as in Fig. 3.

whereas both short (i.e. 16200 bits) and long (i.e. 64800 bits) The two considered approximations are labeled as PW and LUT
LDPC codewords are selected. In the various performance eval- curves and they are shown in dashed lines. Also, min-sum algo-
uation results, which have been obtained by means of computer rithm performance is shown in solid lines, but with some per-
simulations, the coding rate varies from 1/2, 2/3 to 3/4 and the formance degradation compared to EX method (e.g. maximum
modulation format is based on either -ary PSK or -ary 0.8 dB at PER of , depending on the simulation parame-
APSK, e.g. QPSK, 8-PSK, 16-APSK and 32-APSK, respec- ters scenario). It can be seen that for a target PER (or BER) of
tively. Consistent with the previous analysis, an AWGN channel approximately (or ) the loss of LUT with respect to
is considered with a double-sided power spectral density EX method is less than 0.2 dB, whereas the loss of PW with re-
and a maximum of 70 decoding iterations are performed at the spect to EX method is less than 0.1 dB, respectively. Overall, the
decoder. performance evaluation results obtained have shown that such
Performance evaluation results obtained in terms of PER and behavior is independent of the code rate selection, modulation
BER are illustrated in Figs. 3 and 4 for the short LDPC code- format and codeword size. Such observations clearly verify the
word and in Figs. 5 and 6 for the long LDPC codeword, re- robustness of the adopted approximations.
spectively. In all curves denoting the optimum SPA algorithm, In order to have a better view on the performance of different
which uses (1), is labeled as EX and is shown in solid lines. check node update algorithms, Table III depicts the required bit
IEEE TRANSACTIONS ON BROADCASTING, VOL. 54, NO. 1, MARCH 2008 125

Fig. 6. BER performance of DVB-S2 LDPC code, N = 64800 bits. Other system parameters are the same as in Fig. 3.

TABLE III
PERFORMANCE COMPARISON AMONG DIFFERENT DECODING ALGORITHMS FOR THE DVB-S2 LDPC CODE, IN TERMS OF REQUIRED E =N AT PER (OR BER)
OF APPROXIMATELY 10 (OR 10 ). SHORT CODEWORD OF N = 16200 bits, AWGN CHANNEL AND A MAXIMUM OF 70 DECODING ITERATIONS

energy to noise power spectral density ratio at PER (or VI. CONCLUSION
BER) of approximately (or ). In this table, the short The aim of this paper was to study and compare, in a sys-
LDPC codeword is considered together with different modula- tematic manner, different check node update approximation
tion and coding parameters. It can be observed that: (i) the Ja- methods used to decode LDPC codes specified by the DVB-S2
cobian approach performs similarly with the EX method; (ii) standard. Various results have shown small performance degra-
the NMS algorithm performs similarly or slightly better than dation compared to the optimal SPA but significant reduction
the PW approximation, except for the last case (i.e. , in computational complexity was achieved. The different
8-PSK) in which a very small degradation is noticed and (iii) check node update approximations were shown to have robust
the OMS algorithm has very small degradation compared to the performance, i.e. they were independent of the selection of
DVB-S2 modulation and coding parameters. The observed
LUT approximation.
performance/complexity trade-off can be useful in future ap-
Note that all performance evaluation results are comparable plications of DVB-S2 systems, for instance when introducing
with those from [10], although 50 decoding iterations were used mobility aspects, in which computational complexity savings
in this reference. By increasing the number of iterations to 70, could be of significant interest.
only small performance improvement of less than 0.05 dB was
observed. In all performance evaluation results a normalization
REFERENCES
approach to the function was considered, in order to ap-
[1] R. Gallager, Low-Density Parity-Check Codes. Cambridge: MIT
proximate the approach to infinity and prevent decoding over- Press, 1963.
flow [22]. This normalization was based upon the use of clipping [2] R. Tanner, “A recursive approach to low complexity codes,” IEEE
when the input argument was greater than a predefined max- Trans. Inform. Theory, vol. 27, no. 5, pp. 533–547, 1981.
[3] D. J. C. MacKay, “Good error-correcting codes based on very sparse
imum value (e.g. 10), and explains the error floor removal to matrices,” IEEE Trans. Inform. Theory, vol. 45, no. 3, pp. 399–431,
lower values in the corresponding PER/BER curves. 1999.
126 IEEE TRANSACTIONS ON BROADCASTING, VOL. 54, NO. 1, MARCH 2008

[4] M. G. Luby, A. M. Shokrolloahi, M. Mizenmacher, and D. A. [15] J. Chen, A. Dholakia, E. Eleftheriou, M. P. C. Fossorier, and X.-Y.
Spielman, “Improved low-density parity-check codes using irregular Hu, “Reduced-complexity decoding of LDPC codes,” IEEE Trans.
graphs and belief propagation,” in IEEE Inter. Symp. on Inform. Commun., vol. 53, no. 8, pp. 1288–1299, Aug. 2005.
Theory, Cambridge, USA, Aug. 1998, p. 117. [16] E. A. Choi, D. I. Chang, D. G. Oh, and J. W. Jung, “Low computa-
[5] N. Wiberg, “Codes and Decoding on General Graphs,” Ph.D. disserta- tional complexity algorithms of LDPC decoder for DVB-S2 systems,”
tion, Univ. of Linkoping, Linkoping, Oct. 1996. in IEEE Vehic. Techn. Conf. (VTC-Fall), Dallas, USA, Sept. 2005, pp.
[6] T. J. Richardson and R. L. Urbanke, “The capacity of low-density 536–539.
parity-check codes under message-passing decoding,” IEEE Trans. [17] F. Kienle, T. Brack, and N. Wehn, “A synthesizable IP core for DVB-S2
Inform. Theory, vol. 47, no. 2, pp. 599–618, 2001. LDPC code decoding,” in IEEE Design, Automation and Test in Europe
[7] S.-Y. Chung, G. D. Forney, T. J. Richardson, and R. Urbanke, “On (DATE), Munich, Germany, Mar. 2005, pp. 100–105.
the design of low-density parity-check codes within 0.0045 dB of the [18] P. Urard, E. Yeo, L. Paumier, P. Georgelin, T. Michel, V. Lebars, E.
Shannon limit,” IEEE Commun. Lett., vol. 5, no. 2, pp. 58–60, 2001. Lantreibecq, and B. Gupta, “A 135 Mb/s DVB-S2 compliant codec
[8] Digital Video Broadcasting (DVB); Second generation framing struc- based on 64800b LDPC and BCH codes,” in IEEE Int. Solid-State Circ.
ture, channel coding and modulation systems for broadcasting, inter- Conf. (ISSCC), San Francisco, USA, Feb. 2005, pp. 446–609.
active services, news gathering and other broadband satellite applica- [19] J. Dielissen, A. Hekstra, and V. Berg, “Low cost LDPC decoder for
tions, ETSI Std. EN 302.307, June 2004. DVB-S2,” in IEEE Design, Automation and Test in Europe (DATE),
[9] E. Casini, R. D. Gaudenzi, and A. Ginesi, “DVB-S2 modem algorithms Munich, Germany, Mar. 2006, pp. 1–6.
design and performance over typical satellite channel,” Inter. Journ. [20] T. Yokokawa, M. Nakane, and M. Kan, “A low complexity and pro-
Satel. Commun. and Netw. (IJSCN), vol. 22, no. 3, pp. 281–318, May/ grammable encoder architecture of the LDPC codes for DVB-S2,” in
June 2004. 3rd Turbo Coding Conference, Munich, Germany, Apr. 2006.
[10] A. Morello and V. Mignone, “DVB-S2: The second generation stan- [21] G. Albertazzi, S. Cioni, G. E. Corazza, M. Neri, R. Pedone, P. Salmi,
dard for satellite broad-band services,” Proc. IEEE, vol. 94, no. 1, pp. A. Vanelli-Coralli, and M. Villanti, “On the adaptive DVB-S2 physical
210–227, Jan. 2006. layer: Design and performance,” IEEE Trans. Wireless Commun., vol.
[11] IEEE Trans. Broadcast., Special issue on Mobile Multimedia Broad- 12, no. 6, pp. 62–68, 2005.
casting, vol. 53, no. 1, Mar. 2007. [22] S. Papaharalabos, P. Sweeney, B. G. Evans, P. T. Mathiopoulos, G. Al-
[12] J. Song, Z. Yang, L. Yang, K. Gong, C. Pan, J. Wang, and Y. Wu, “Tech- bertazzi, A. Vanelli-Coralli, and G. E. Corazza, “Modified sum-product
nical review on Chinese digital terrestrial television broadcasting stan- algorithms for decoding low-density parity-check codes,” IET Proc.
dard and measurements on some working modes,” IEEE Trans. Broad- Commun., vol. 1, no. 3, pp. 294–300, June 2007.
cast., vol. 53, no. 1, pp. 1–7, Mar. 2007. [23] M. Eroz, F.-W. Sun, and L.-N. Lee, “DVB-S2 low density parity check
[13] W. Zhang, Y. Guan, W. Liang, D. He, F. Ju, and J. Sun, “An introduc- codes with near Shannon limit performance,” Inter. Journ. Satel.
tion of the Chinese DTTB standard and analysis of the PN595 working Commun. and Netw. (IJSCN), vol. 22, no. 3, pp. 269–279, May/June
modes,” IEEE Trans. Broadcast., vol. 53, no. 1, pp. 8–13, Mar. 2007. 2004.
[14] S. Scalise, G. E. Corazza, C. P. Niebla, P. Chan, G. Giambene, F. [24] P. Robertson, E. Villebrun, and P. Hoeher, “A comparison of optimal
Hu, A. Vanelli-Coralli, and M. A. Vázquez-Castro, “Towards the revi- and sub-optimal MAP decoding algorithms operating in the log do-
sion of DVB-S2/RCS standard for the full support of mobility,” IEEE main,” in IEEE Int. Conf. Commun. (ICC), Seattle, USA, June 1995,
Commun. Society, SSC Newsletter, vol. 16, no. 2, pp. 6–8, Nov. 2006. pp. 1009–1013.

You might also like