Fpga implementation of linear ldpc encoder

IJRET: International Journal of Research in Engineering and Technology eISSN: 2319-1163 | pISSN: 2321-7308
__________________________________________________________________________________________
Volume: 02 Issue: 11 | Nov-2013, Available @ http://www.ijret.org 442
FPGA IMPLEMENTATION OF LINEAR LDPC ENCODER
Chetna N. Kharkar1
, M. M. Jadhav2
, A. M. Sapkal3
1
Sr.Design Engineer, Qualitat Systems,Pune,Maharashtra,India,chetnakharkar@gmail.com
2
Asso. Professor, E&TC, Sinhagad college of Engineering, Pune, Maharashtra,India,makj123@yahoo.com
3
HOD, E&TC, Government College of Engineering, Pune, Maharashtra, India, hod.extc@coep.ac.in
Abstract
In this paper, a FPGA implementation of linear time LDPC encoder is presented. This encoder implementation can handle large size
of input message. Linear Time encoder hardware architecture reduces the Complexity and area of encoder than generator matrix
based encoder techniques. This encoder is simulated on different platform which includes Matlab & High level languages for 1/2 rate
& up to 4096 code length. FPGA implementation of the encoder is done on Xilinx Spartan 3E Starter Kit. The result shows the speed
& area comparison for different FPGA platform.
Keywords— LDPC codes, dual-diagonal, Linear encoding, Generator matrix complexity, FPGA Implementation
---------------------------------------------------------------------***---------------------------------------------------------------------
1. INTRODUCTION
As their name suggests, LDPC codes are block codes with
parity-check matrices that contain only a very small number of
non-zero entries proposed by Gallager in 1962 [1] and has
gained popularity due to their capacity-approaching error
correcting performance [2].
In LDPC codes sparseness of H guarantees both a decoding
complexity and minimum distance which increases only
linearly with the code length, however, finding a sparse parity-
check matrix for an existing code is not practical. Instead
LDPC codes are designed by constructing a sparse parity-
check matrix first and then determining a generator matrix for
the code afterwards.
In order to reduce encoding complexity, LDPC codes with
dual diagonal structure is adopted by the latest next-generation
wireless LAN standard, IEEE 802.11n [3]. The LDPC
encoding algorithm used is near-linear time proposed by [4] &
[5].An LDPC code parity-check matrix is called (wc,wr)-
regular if each code bit is contained in a fixed number, wc, of
parity checks and each parity-check equation contains a fixed
number, wr, of code bits. An efficient encoding algorithm [6]
is used to reduce the encoding complexity.
In this paper we have implemented the low complexity
Encoder algorithm on hardware platform on Xilinx Spartan 3E
FPGA & simulated using Matlab 2012, Modelsim & c code.
The Synthesis results shows the area & speed comparison on
different FPGA platform. The encoded codeword is decoded
using belief propagation algorithm [7] & results are verified
using Matlab program.
2. LDPC CONSTRUCTION
The construction of binary LDPC codes involves assigning a
small number of the values in an all-zero matrix to be 1 so
that the rows and columns have the required degree
distribution.
The original LDPC codes presented by Gallager are regular
and defined by a banded structure in H. The rows of
Gallager‟s parity-check matrices are divided into wc sets
with M/wc rows in each set. The first set of rows contains wr
consecutive ones ordered from left to right across the
columns. (i.e. for i ≤ M/wc, the i-th row has non zero entries
in the ((i − 1)K + 1)-th to i-th columns). Every other set of
rows is a randomly chosen column permutation of this first
set. Consequently every column of H has a „1‟ entry once in
every one of the wc sets. Since LDPC codes are often
constructed pseudo-randomly we often talk about the set (or
ensemble) of all possible codes with certain parameters (for
example a certain degree distribution) rather than about a
particular choice of parity-check matrix with those
parameters. LDPC codes are often represented in graphical
form by a Tanner graph.
The Tanner graph as shows in figure-1, consists of two sets
of vertices: n vertices for the code word bits (called bit
nodes), and m vertices for the parity-check equations (called
check nodes). An edge joins a bit node to a check node if
that bit is included in the corresponding parity-check
equation and so the number of edges in the Tanner graph is
equal to the number of ones in the parity-check matrix.

__________________________________________________________________________________________
Fig 1: The Tanner graph representation of the parity-check a
6-cycle is shown in bold.
A cycle in a Tanner graph is a sequence of connected
vertices which start and end at the same vertex in the graph,
and which contain other vertices no more than once. The
length of a cycle is the number of edges it contains, and the
girth of a graph is the size of its smallest cycle. The Mackay
Neal construction method for LDPC codes can be adapted to
avoid cycles of length 4, called 4-cycles, by checking each
pair of columns in H to see if they overlap in two places.
The construction of 4-cycle free codes using this method is
given in Algorithm 1. Input is the code length n, rate r, and
column and row degree distributions v and h. The vector α is
a length n vector which contains an entry i for each column
in H of weight i and the vector β is a length m vector which
contains an entry i for each row in H of weight i.
Algorithm 1: H Matrix Generation
Procedure
MNCONSTRUCTION (n, r, v, h) ⊲Required length, rate
and degree distributions
H = all zero n(1 − r) × n matrix ⊲ Initialization
α = [];
for i = 1 : max(v) do
for j = 1 : vi × n do
α = [α, i]
end for
end for
β = []
for i = 1 : max(h) do
for j = 1 : hi × m do
β = [β, i]
end for
end for
for i = 1 : n do ⊲ Construction
c = random subset of β, of size αi
for j = 1 : αi do
H(cj , i) = 1
end for
α = α − c
end for
repeat
for i = 1 : n − 1 do ⊲ Remove 4-cycles
for j = i + 1 : n do
if |H(:, i)
S
H(:, j)| > 1 then
permute the entries in the j-th column
end if
end for
end for
until cycles removed
end procedure
3. ENCODING USING GENERATOR MATRIX
For a linear block code, the sum of any two code words results
in another code word. LDPC code construction is also done in
similar way of linear block code. From a given parity check
matrix, H, a generator matrix, G is derived. Data, m = m1,
m2…..mn is encoded by multiplying it with the generator
matrix, c = mG where m is a string of information bits. It has
to be noted that putting H in systematic form, H= [P T
| IM], no
longer has fixed column or row weights and P is very likely to
be dense. The denseness of P determines the encoder
computational complexity. A dense generator matrix requires
a large number of operations when doing the matrix
multiplication with the data to be sent. The encoding
complexity could be reduced for some codes by parity check
matrix pre-processing. An efficient encoding technique has
been developed to reduce the encoding complexity by
rearranging the parity check matrix before encoding. The
encoding complexity also depends on the structure of the code
The construction of LDPC codes is categorized mainly into
two: Random constructions and structured constructions. The
type of construction is determined by the connections between
check nodes and variable nodes in Tanner graph. Each type of
constructions has their advantages over the other. Random
constructions refer to the unstructured row-column
connections in the parity check matrix with no predefined
pattern. Random codes have better performance compared to
structured codes in case of long codes. They are used in cases
we want to increase the girth or rate of a given size. But longer
length random LDPC codes require large memory storage in
practical implementation which affects the computational
efficiency of the code. The uncertainty of guaranteeing an
asymptotically optimum performance in random constructions
leads to the use of structured construction of LDPC codes.
Structured construction method put constraints on row –
column connections to get a desired or predefined connection
pattern that is easier to implement in hardware.
3.1 LDPC Encoding Example
0 1 0 1 1 0 0 1
H = 1 1 1 0 0 1 0 0
0 0 1 0 0 1 1 1
1 0 0 1 1 0 1 0

__________________________________________________________________________________________
The NMH  parity check matrix defines a rate NKR  ,
 KN, code where MNK  .
Code word is said to be valid if it satisfies the syndrome
calculation:
0.  T
Hcz
We can generate the code word in by multiplying message m
with generator matrix G
c = m.G
We can obtain the generator matrix G from parity check
matrix H by:
1. Arranging the parity check matrix in systematic form
using row and column operations
 KMMsys PIH 












110100
111010
101001
sysH
2. Rearranging the systematic parity check matrix
 K
T
MK IPG  ,











100111
010110
001011
G
3. We can verify our results as 0. T
HG
4. LINEAR-TIME ENCODING FOR LDPC CODES
Instead of finding a generator matrix for H, an LDPC code
can be encoded using the parity-check matrix directly by
transforming it into upper triangular form and using back
substitution. The idea is to do as much of the transformation as
possible using only row and column permutations so as to
keep as much of H as possible sparse.
Firstly, using only row and column permutations, the parity-
check matrix is put into approximate lower triangular form:
Ht =
Where the matrix T is a lower triangular matrix (that is T has
ones on the diagonal from left to right and all entries above the
diagonal zero) of size
(m − g) × (m − g)
If Ht is full rank the matrix B is size
m − g × g
And A is size
m − g × k
The g rows of H left in C, D, and E are called the gap of the
approximate representation and the smaller g the lower the
encoding complexity for the LDPC code.
Step 1
Instead of putting H into reduced row-echelon form we put it
into approximate lower triangular form using only row and
column swaps. For this H we swap the 2-nd and 3-rd rows and
6-th and 10-th columns to obtain:
Ht =
1 1 0 1 1 0 0 1 0 0
0 0 0 1 0 1 0 1 1 0
0 1 1 0 1 0 1 0 0 1
1 1 0 0 0 0 1 0 1 1
0 0 1 0 0 1 0 1 0 1
With a gap of two
Once in upper triangular format, Gauss-Jordan elimination is
applied to clear E which is equivalent to multiplying Ht by
,
To give
where
And
From Step 1
A B T
C D E
Im-g 0
-ET-1
Ig
Im-g 0
H= -ET-1
Ig̃
C = -ET-1
A+C̃
D = -ET-1
B+D̃
A B T
Ht = C D 0̃ ̃

__________________________________________________________________________________________
Step 2
And
To give
When applying Gauss-Jordan elimination to clear E only C
and D are affected, the rest of the parity-check matrix remains
sparse. Finally, to encode using H the code word
c = [c1c2, . . . , cn]
Is divided into three parts,
c = [u, p1, p2],
Where
u = [u1, u2, . . . , uk]
Is the k-bit message
p1 = [p11 , p12 , . . . , p1g ],
Holds the first g parity bits and
p2 = [p21 , p22 , . . . , p2m−g ]
Holds the remaining parity bits
The code word
c = [u, p1, p2]
Must satisfy the parity-check equation c H T
=0 and so
Au + Bp1 + Tp2 = 0, ---- (1)
And
Cu + Dp1 + 0p2 = 0. ---- (2)
Since E has been cleared, the parity bits in p1 depend only on
the message bits, and so can be calculated independently of
the parity bits in p2. If D is invertible, p1 can be found from
Equation (2)
p1 = D−1 Cu. ---- (3)
If D is not invertible the columns of H can be permuted until it
is. By keeping g as small as possible the added complexity
burden of the matrix multiplication in Equation (3), which is
(g2), is kept low. Once p1 is known p2 can be found from
Equation (1)
p2 = −T−1(Au + Bp1), ---- (4)
Where the sparseness of A, B and T can be employed to keep
the complexity of this operation low and, as T is upper
triangular, p2 can be found using back substitution.
From Step 2 we partition the length 10 codeword c = [c1, c2, .
. . , c10] as c = [u, p1, p2] where p1 = [c6, c7] and p2 = [c8,
c9, c10]. The parity bits in p1 are calculated from the message
using Equation 3
Step 3
As T is upper-triangular the bits in p2 can then be calculated
using back substitution
and the code word is c =
Again column permutations were used to obtain Ht from H
and so either Ht, or H with the same column permutation
applied, will be used at the decoder.
5. ENCODER DESIGN & IMPLEMENTATION
Hardware implementation of Encoder is done on Xilinx
Spartan 3E FPGA starter kit. Figure 2 shows the flow diagram
for encoder implementation. We have implemented the ½ rate
encoder for different matrix size 4X8, 16X32, 32X64, 64X128
1 0 0
T-1
= 1 1 0
0 0 1
̃
̃
̃
̃
̃ ̃
̃ ̃
̃
̃
p21 = u1 u2 u4 u5= 1 1 0 1=1
p22 = u4 p11 p21=0 1 1 =0
p23= u2 u3 u5 p12 = 1 0 1 0 = 0
1 1 0 0 1 1 0 1 0 0
̃
Im-g 0
-ET-1
Ig =
1 0 0 0 0
0 1 0 0 0
0 0 1 0 0 ,
1 1 1 1 0
1 0 1 0 1
1
1
1 0 0 1 1 0 0 0 = 1 0
1 1 1 0 0 1 0 0
1
p1= D-1
C =̃ ̃
̃H =
1 1 0 1 1 0 0 1 0 0
0 0 0 1 0 1 0 1 1 0
0 1 1 0 1 0 1 0 0 1
0 1 1 0 0 1 0 0 0 0
1 0 0 1 0 1 1 0 0 0

__________________________________________________________________________________________
on FPGA.We have used the Xilinx Vivado high level
synthesis tool for design of FPGA based encoder. This tool
supports the High level synthesis feature, using this feature we
have done synthesis of our high level program. The Encoder
design is synthesized on different FPGA & results are
compared in terms of area & speed.
Fig-2 Encoder Implementation flow
6. IMPLEMENTATION RESULTS
We have simulated the ldpc encoder & log domain decoder
algorithm in Matlab & results are verified both in simulation
& implementation. Figure-4 shows Matlab simulation results.
We have implemented linear-time Encoder for LDPC codes.
This algorithm is implemented on Xilinx Spartan 3E board
using ISE 13.1 & Xilinx High Level synthesis vivado HLS
tool.
The synthesis results for Spartan 3E FPGA are shown in
Figure 3.
Fig -3 Device Utilization for Spartan 3E FPGA
Table-1 Comparison of Area & speed
Selected Device Number of Slice
Registers
Clock
Frequency
3s500efg320-4 1506 (out of 4656 ) 71.782MHz
6slx4tqg144-3 2321 (out of 4800 ) 117.427MHz
7a30tcsg324-3 1918 ( out of 42000) 187.337MHz
Encoder performance is verified on different FPGA platform
table 1 shows the comparison of area & speed, from the table
it is clear that Xilinx 7a30tcsg device Supports
Faster design speed.
Fig-4 Result of Encoding & Decoding in Matlab simulation
CONCLUSIONS
We have implemented the linear time encoder in simulation &
synthesis is done using Xilinx Tool. Xilinx Spartan 3E starter
board is used for hardware implementation. The algorithm
accepts the inputs as a input Message, H-matrix size &
generates the Encoded codeword as a output.
This algorithm we have simulated on various platform
including Matlab, C code, ISE13.2 & Modelsim .Building the
Encoded codeword using Generator matrix is complicated for
large size of parity check matrix .This Linear Time encoder
algorithm provides an alternative for generator matrix creation
& suitable for large parity check matrix .We have simulated &
implemented LDPC encoder algorithm for smaller as well as
larger codeword.
Parity Check
Matrix H
Perform Row &
column swapping
Check H for
lower
Triangular
Apply Gauss-Jordan
to clear E
No
Yes
If E matrix
clear
Yes
No
New H matrix

__________________________________________________________________________________________
REFERENCES
[1]. R. G. Gallager, “Low density parity check codes,”IRE
Trans. Inform. Theory, vol. IT-8, pp. 21–28, Jan. 1962.
[2]. D. J. C. MacKay and R. M. Neal, “Near Shannon limit
performance of low density parity check codes,” Electron.
Lett., vol. 32, p. 1645, 1996.
[3]. T.J. Richardson and R.L. Urbanke, “Efficient encoding of
low density parity-check codes,” IEEE Trans. Inform. Theory,
vol.47, no.2 pp. 638- 656, Feb. 2001
[4]. D. J. C. MacKay, “Good error-correcting codes based on
very sparse matrices,”IEEE Trans. Inform. Theory, vol. 45, no.
2, pp. 399–431, March
1999.
[5]. C. Yoon, J.-E. Oh, M. Cheong Cheong, and S.-K. Lee, “A
hardware efficient LDPC encoding scheme for exploiting
decoder structure and resources,” IEEE Vehicular Technology
Conference (VTC2007-Spring), pp. 2445-2449, April 2007.
[6]. Susmitha Remmanapudi,and Balaji Bandaru, “An FPGA
Implementation of low density Parity-check codes
construction & Decoding” Devices ,Circuits & Systems
International conference, pp .216-220,April 2012.
BIOGRAPHIES
Chetna N. Kharkar, She has completed her
Graduation in Electronics &
telecommunication Engineering. Currently
working as a Sr. Design Engineer at Qualitat
systems, Pune (India).She has a 6+ years of
experience in Embedded domain. Perusing
Master of Engineering from Sinhagad college of Engineering,
Pune
M. M. Jadhav, he has Completed his
graduation & Post graduation from
Government college of Engineering, Pune.
Presently perusing PHD from Government
college of Engineering, Pune in the field of
communication He has guided graduates & more than ten Post
graduates students, His area of Interest is communication
system
Dr. A. M. Sapkal, HOD of E&TC
Department at government college of
Engineering, Pune.He has 23 years of
experience in teaching, 3 years Industrial &
14 years of Research experience. He has
guided More than 100 post graduates
students. He is a Chairman of Central Library of College of
Engineering Pune.He is a member of THE IET U.K., IEEE,
MIETE India.

Fpga implementation of linear ldpc encoder

More Related Content

What's hot (19)

Viewers also liked (16)

Similar to Fpga implementation of linear ldpc encoder (20)

More from eSAT Journals (20)

Recently uploaded (20)

Fpga implementation of linear ldpc encoder