Page 1 of 6
Journal for Studies in Management and Planning
Available at http://edupediapublications.org/journals/index.php/JSMaP/
ISSN: 2395-0463
Volume 04 Issue 03
March 2018
Available online: http://edupediapublications.org/journals/index.php/JSMaP/ P a g e | 222
Robust Wireless Image Transmission Using
Asymmeteric Turbo Codes
G. Vijay Kumar, K. Bhavana, N. Sai Teja, G.Veera Venkata Lakshmi
Assistant Professor, Dept. Of ECE, Andhra Loyola Institute of Engineering and Technology
Vijayawada, Andhra Pradesh, India.
Btech scholars, Dept. Of ece, Andhra Loyola Institute of Engineering and Technology
Vijayawada, Andhra Pradesh, India.
vijay.gayaala@gamil.com, kbhavana997@gmail.com, saiteja0067@gamil.com, lakshmaigurrana319@gmail.com
Abstract— Today's world thrives on information exchange.
Hence the need of the day is the information be protected
well enough to the transmitted over a noisy environment.
This is achieved by adding redundant bits to the
information bit streams. If the purpose of adding
redundancy bits is just to detect error and inform the
sender to retransmit the information. Forward error
correction (FEC) is another way of adding redundancy to
the information bit stream. So, error can be detected and
corrected by preventing needed retransmission.
Turbo codes is a very powerful error correction
technique that has made tremendous impact on channel
coding in last few years. Turbo code bit error rate drops
very rapidly with increasing Eb/N0 values. It achieves 10^-
5 BER with Recursive Systematic Convolution encoder
.The iterative decoding mechanism, RSC and use of
interleave are the characteristics features of turbo codes.
That it enhance data transmission efficiency in digital
communication system. Turbo codes play a major role in
multimedia services in mobile phones. The performance of
turbo codes is superior with a little Eb/No.
The original JPEG (Joint Photography Expert
Group) image is encoded using turbo codes and subjected
to additive white Gaussian noise. In this random
interleaver is used and MAP decoding algorithm is used.
We can almost retrieve original image by number of
iterations by iterative decoder. As the number of iterations
increases the noise in image removed.
Keywords: Turbo Coding, Forward error
correction, Interleaving, puncturing, Iterative decoding,
MAP decoding.
1. INTRODUCTION
Turbo Code proposed in 1993 by Berrou et al, is
known for excellent coding gain. It provides the error free
communication near to Shannon Limit at great extent. Due to
many research efforts of the turbo coding community, it is
used in standardized system such as third-generation (3G)
mobile radio system and so many other emerging wireless
Applications.
Basically, the Turbo code can be classified into two
types based on their generator polynomial structures. The
component with identical encoders is basically known as
symmetric turbo codes, otherwise asymmetric turbo code. The
parallel concatenated turbo codes can assumes identical
component code, as in the Symmetric turbo codes, have either
a good ―waterfall‖ Bit Error Rate (BER) performance or a
good ―error floor‖ BER performance but not both. Since, the
asymmetric turbo code uses non identical component codes
and can be designed with proper selection of weight
configuration for better BER performance. In this paper,
several new classes of asymmetric turbo codes are introduced
which improves the performance compared to the original
turbo codes (symmetric) over the entire range of signal to
noise ratios. A practical setup with symmetric and asymmetric
turbo codes is described and the performance results are
discussed.
Figure.1. Block diagram of turbo code encoder.
2. ASYMMETRIC TURBO CODE
The turbo code with non-identical component
encoders is known as Asymmetric turbo codes. The BER
curve of a turbo code is divided into two region i.e. ―waterfall‖
region and ―error floor‖ region. ―Waterfall‖ region is given as
a steep slope for a long block of information bits and appears
at a smaller SNR’s but ―error floor‖ region appears at higher
SNR’s and has a shallower slope due to code words of small
weight. So, by using symmetric turbo codes we can’t get the
better BER performance simultaneously for both waterfalls as
well as error floor region. In that respect asymmetric turbo
code satisfy the requirements for both the regions provided the
selection of component encoders are proper.
A. Asymmetric Turbo Encoders
The asymmetric turbo code, like symmetric code has
two un-identical recursive systematic convolutional (RSC)
codes which generate the systematic codeword that consist of
parity bit and information bit. The block diagram of turbo
encoders are shown in figure 1. Two components encoders are
separated by an interleaver.
Page 2 of 6
JOURNAL FOR STUDIES IN MANAGEMENT AND PLANNING
AVAILABLE AT
HTTP://EDUPEDIAPUBLICATIONS.ORG/JOURNALS/INDEX.PHP/JSMAP/
ISSN: 2395-0463
VOLUME 04 ISSUE 03
MARCH 2018
Available online: http://edupediapublications.org/journals/index.php/JSMaP/ P a g e | 223
In fig.1, we can see that there are three outputs,
systematic output (v0), and two recursive convolutional
sequence output (v1 and v2). Two parallel concatenated RSC
encoders are joint with an interleaver. The simple structure of
turbo encoders with code rate 1/2, constraint length 3 with un- identical components shown in figure 1.
B. Asymmetric Turbo Decoders
In this case also we can use similar decoding
algorithms which are applicable for symmetric turbo decoders
like Maximum-a-posteriori (MAP), Logarithmic Maximum-a
posteriori (Log-MAP), Maximum Logarithmic Maximum-a
posteriori (Max Log-MAP) and Soft output Viterbi decoding
(SOVA). However we use un-identical component code in the
Corresponding turbo decoders.
The MAP algorithm is the optimum decoding
technique but the complexity is high. It is used to determine
the most probable information bit that was transmitted but the
SOVA is used for most probable information sequence that
was transmitted. In Max Log-MAP the values and operation
are easier to implement due to logarithmic domain but Log- MAP avoid the approximation as in Max Log-MAP. Hence,
we used Log-MAP decoding algorithm for performance
evaluation with low computational cost without much
compromise in the BER performance. The block diagram of
Log MAP turbo decoder is shown in figure 2.
Figure.2.Block Diagram of Turbo decoder
The improvement in the error floor region can be
done by serially concatenated turbo code or a parallel
concatenated code of primitive components which have worse
performance than original Berrou code in waterfall region. In
asymmetric turbo code we consider the performance
characteristic in both region i.e. in ―waterfall‖ region as well
as ―error floor‖ region. In this paper we reduce the flattening
of the ―error floor‖ curve by applying asymmetric turbo code.
The asymmetric turbo code with encoder with (7,5) and
(15,17) component codes taking half code rate is shown in
figure 3. Here, we can see that the two component codes are
not identical, so it can be treated as asymmetric turbo code.
The generated polynomial of the components codes are
constructed with mixed type of the primitive polynomial and
prime polynomial.
Figure.3.Block Diagram of Asymmetric Turbo Encoder
3. THE EFFECT OF VARIOUS CODEC PARAMETERS
IN ASYMMETRIC TURBO CODE
There are many parameters, which affect the
performance of asymmetric turbo codes. The various
simulation results for asymmetric turbo code by using Binary
Phase Shift Keying (BPSK) over Additive White Gaussian
Noise (AWGN) channels are presented in this section. The
parameters which affect the performance are as follows:-
a. The number of decoding iteration
b. Puncturing (or code rate)
c. Frame-Length
d. Component codes
e. Constraint length
The parameters which we have used in our simulation
are shown in Table I. Before going through the various results
for different parameters, we have tested and verified the
simulation model of asymmetric turbo code by substituting
generator polynomials as g1=g2= (15, 13) than comparing the
result with symmetric turbo code for g0= (15, 13)
Page 3 of 6
JOURNAL FOR STUDIES IN MANAGEMENT AND PLANNING
AVAILABLE AT
HTTP://EDUPEDIAPUBLICATIONS.ORG/JOURNALS/INDEX.PHP/JSMAP/
ISSN: 2395-0463
VOLUME 04 ISSUE 03
MARCH 2018
Available online: http://edupediapublications.org/journals/index.php/JSMaP/ P a g e | 224
The curves shown in figure 4 verify the asymmetric
turbo codes simulation model. It is found that the result is
matching with symmetric turbo codes BER performance, as
expected. Mainly, the generated polynomials are optimum in
terms of maximizing the minimum free distance of the
components codes [9]. Most of the results in this paper for half
code rate and also the decoding technique used are Log MAP
decoder. All simulation results are taken over an AWGN
channel with BPSK modulation.
A. The Effect of Number of iterations used
The performance of an asymmetric turbo code using
Log-MAP algorithm with different decoding iteration is shown
in figure 5. The generated polynomial used for the encoders
are taken as (7, 5) and (5, 7). It can be seen from the above
figure that the performance is nearly same as the encoded bits
at low Eb/N0 but at high Eb/N0 the BER performance is
improved after one iteration. When we increase the iteration
like 2, 4, 6 and 8 then we get the better performance
progressively. But after 6 iteration, there are a little
improvement in performance approximately less than 0.1dB,
so we use only 8 iteration due to complexity reason because as
we increase the iteration more accurate the result, so more
complexity.
Figure 5: Performance using different number of iteration
B. The Effect of Puncturing or different code rate
Half of the parity bits from each component encoders
are punctured when we use the half-rate code. But it is
possible to avoid the puncturing and transmit all the parity bits
through both the components encoder with one third code rate.
Hence, the figure 6, shows the performance of BER taking
parameters from table I, but the code rate is different i.e. half
and one third. Like symmetric turbo code, the effect of
puncturing in asymmetric turbo is also similar and effect
reflected in the figure 6 which shows the performance graph
for rate one-third is better than the rate one-half.
Figure 6: performance using half and one third code rate
C. The Effect of Frame Length
The BER performance is better as we increase the
frame length. Since, the analysis of associated theoretical
performance limits as a function of the coded frame length is
already given by Dolinar et a in [10].
So, a large number of frame length is an unacceptable in real
time performance because of the delay in transmission. In
speech transmission we use 169 bit code while in video
transmission we use 1000 bit code .So as we increase the
frame length we don’t get the real time transmission however
it would be useful in data or non-real time transmission.
D. The Effect of Components codes
The generator polynomial is also the important
parameter used in the component codes. Figure 8, shows the
different generated polynomials which affect the performance
