This full-text paper was peer-reviewed and accepted to be presented at the IEEE WiSPNET 2017 conference.

Reducing Bit Error Rate Using CRC

Verification in Turbo Codes
G. Sowndharya1 and A. Vasuki2
Department of Electronics and Communication Engineering, Kumaraguru College of technology, Coimbatore-641049, India
Email: 1 govindhansowndharya@gmail.com 2 vasuki.a.ece@kct.ac.in

Abstract—In this paper, MAX Log MAP decoding algorithm

is applied in turbo decoding process to reduce the computa-
tional complexity in LTE standard. In addition, an algorithm is
proposed to reduce the bit error rate in turbo codes by apply-
ing Cyclic Redundancy Check and Flip and Check algorithm.
The data bits have been transmitted through Rayleigh fading
channel and the performance of proposed decoding algorithm is
estimated. The performance of this algorithm has been analyzed
using Bit Error Rate by varying the Signal to Noise Ratio
(Eb/No). The code rate is 1/3, number of iterations has been
increased from 1 to 8 and the size of the block is varied from
40 to 6144. The algorithm has been implemented and simulated
using MATLAB software.
Index Terms—Turbo codes, MAP, MAX Log MAP, CRC, Flip
& Check algorithm, LTE standard.

I. I NTRODUCTION
In a digital communication systems, information is
represented as a sequence of binary data. While the data is
transmitted through the channel, error information may occur
it will affect the transmitted data due to interference. In order
to minimize those errors it requires error detection and cor-
rection technique. Channel coding is one of the techniques for
detecting and correcting error occurrence in the received data
bits. The detected errors can be corrected by using Forward
Error Correction technique. The Forward Error Correction
technique is classified into various types such as block codes,
convolutional codes, Low density parity check codes and
Turbo codes. Among those codes turbo codes is the powerful
error detecting and correcting codes. It was first developed
by Berrou in 1993 [1], [2]. The performance of these codes
was nearer to the theoretical Shannon limit value and also
provides low computational complexity by using appropriate
decoding algorithm [3] in the turbo decoding process. It is
used for many wireless communications, both terrestrial and
space applications such as the Long Term Evaluation, universal
mobile Telecommunication System, Digital Video Broadcast
and Worldwide Interoperability for microwave access stan-
dards [4].
Turbo encoder consists of parallel connection of recursive
systematic convolutional (RSC) encoders; each convolutional
encoder produces a different encoded output. In turbo codes,
the interleaver is the important block to improve the perfor-
mance which is connected between the two encoders. It is
mainly used for scrambling the input data bits. Because of the
presence of interleaver, the second encoder produces different
output. This parallel concatenation gives lower convergence
Fig. 1. LTE turbo encoder.
threshold than obtained with a serial concatenation of convo-
lutional encoder. The turbo decoding is an iterative process
which is mainly based on extrinsic information The extrinsic
information is exchanged between the two soft input soft-
output (SISO) decoders [5]. In this paper, we propose a method
to reduce the bit error rate by using cyclic redundancy check
algorithm and flip and check algorithm during the iterative
process. In this algorithm, first iterates minimum times for
identifying the least reliable bits among the input data set.
After identifying the reliable bits, CRC algorithm is start to
check the input data set.
The rest of this paper is organized as follows. Section II
describes the LTE turbo encoder with interleaver de-
sign. Section III explains about turbo decoding principles.
InSection IV brief descriptions of the proposed CRC and
FC algorithms. Section V simulated results are discussed. In
Section VI final conclusion of this work.
II. LTE T URBO E NCODER
Turbo encoder consists of two constituent encoder and
one internal interleaver module. Turbo encoding scheme is
mainly used for 3rd Generation partnership project (3GPP)
specification. It is explained in Fig. 1. The data block Ck is the
input of the first constituent encoder. The constituent encoder
process the data block Ck resulting parity bit Zk , while the
interleaver scramble the input data block is represented as
Ck . The data block Ck is processed by the second interleaver
produces the parity bit Zk .

978-1-5090-4442-9/17/$31.00 2017
c IEEE 627
 This full-text paper was peer-reviewed and accepted to be presented at the IEEE WiSPNET 2017 conference.

TABLE I
C OMPUTATIONAL C OMPLEXITY C OMPARISON .

Operation MAP algorithm MAX-Log MAP

algorithm
Fig. 2. Interleaver design. Maximization 2M − 1 = 5 5M − 2 = 13
Number of addition required 4M = 12 10M = 30
Number of multiplication required 10M = 30 0
Total operation 14M = 42 10M − 2 = 28

end of decoding process. It provides accurate value compared

to hard decision. There is no round off in soft decision. It is
based on the minimum Euclidean distance of the code word. It
helps to improve the performance in turbo codes. The number
of iteration (IT C ) is repeated until a convergence take place or
reached the predetermined value. In turbo decoding process,
both SISO decoders estimate the output information. Due to
the presence of interleaver and de-interleaver, the estimation
takes place by different order [11–13].
Decoding procedure is mainly based on trellis method
Fig. 3. LTE turbo decoder. to calculate the output values by two metrics. The trellis
procedure traces backward direction to find the backward state
Combining all the parity bits, output of the encoder se- metric and trace forward direction to find the forward state
quence is Xk , Xk , Ck , Ck , Zk , Zk . Interleaver is used to metric. The branch metric and state metric are used to find
separate the two encoders. It is used to find low weight code the Log likelihood ratio (LLR).
word of the data. Size of the interleaver is also one of the  
p(uk = +1/y)
important aspects to improve the BER performance in turbo L(uk /y) = (2)
p(uk = −1/y)
codes. Interleaver block is used in encoder side for scramble 
the input data bits, at the decoder side de-interleaver is used αk (s) = γk (s , s) · αk−1 (s ) (3)
to descramble the received data bits. The random interleaver 
 
βk−1 (s) = γk (s , s) · αk−1 (s ) (4)
is chosen for this system which shows in Fig. 2. The input
data should be stored in natural order and then read the data L(uk /y)
  
in different manner [6–9]. The interleaver values has been (s , s) ⇒ uk = +1 · γk (s , s) · βk (s)
obtained from Eq. (1). = ln 
(s , s) ⇒ uk = −1 · αk−1 (s ) · γk (s , s) · βk (s)
(5)
Π(i) = {[(f 1 + f 2 · i) mod k] · i} mod k. (1)
α, β, γ represents Forward state metric, Backward state metric
The first step is find (f1 +f2 · i) mod k data values. Those and Branch metric values. In MAP decoding algorithm, the
obtained values are multiplied by natural order i, after finding LLR is calculated by using Eqs. (2), (3), (4) and (5). In MAX
the index value i take modulo function of that value. It takes Log MAP decoding algorithm, the metrics is calculated by
one clock period. At the second stage, the index value should using Eqs. (6), (7), (8) and (9).
multiplied with the resultant data and taking modulus of the
final value. The modulus operation reduces the time period for Ak (s) = ln(αk (s)) (6)
processing the data. At the end of this process obtained the Bk (s) = ln(βk (s)) (7)
interleaved index value [10].  
Γk (s , s) = ln(Γk (s , s)) (8)
III. LTE T URBO D ECODER L(uk ) = max(s ,s) ⇒ uk
Turbo decoding is an iterative procedure which is used to = +1 (Ak−1 (s ) + Γk (s , s) + Bk (s))
decode the original input data by using parallel concatenation − max(s ,s) ⇒ uk = −1(Ak−1 (s ) + Γk (s , s) + Bk (s)).
of decoders. It consists of two soft input and soft output (9)
decoders, interleaver, de-interleaver. The turbo decoder is In MAX Log MAP algorithm, LLR values are calculated
shown in Fig. 3. The first SISO decoder receives intrinsic using Eq. (9). This paper describes comparison between MAP
information from the encoder block (Xk ) and it produces and MAX Log MAP decoding algorithm.
extrinsic information (E1) as the input of second encoder
 The MAP decoding algorithm requires more number of
receives extrinsic information through the interleaver ( ) or
−1 addition and multiplication; it increases the computational
de-interleaver ( ). The second decoder output is used as a complexity but reduces the bit error rate. Table I shows
intrinsic information (E2) through dinterleaver to the first de- the computational complexity comparison of turbo decoding
coder in next iteration. Turbo decoder use soft decision at the algorithm. MAP algorithm is twice as complex as MAX Log

628
 This full-text paper was peer-reviewed and accepted to be presented at the IEEE WiSPNET 2017 conference.

Fig. 4. Block diagram of CRC.

MAP algorithm for storage memory M = 3. MAX Log MAP

algorithm requires less number of addition and multiplication
because it is work in log domain. It has less computational
complexity compared to other decoding algorithm [14–19].
IV. C YCLIC R EDUNDANCY C HECK A LGORITHM Fig. 5. Flow chart of CRC algorithm.
CRC is an error detection technique in digital data trans-
mission. In this method, the original message contains (k-
bit); the transmitter generates n-bits, those bits are called as
Frame Check Sequence. In the transmitter side, the message
sequence consists of k +n bits that should be divided by some
CRC polynomial based on the application. At the receiver
side, incoming data should be divided by the same CRC
polynomial, if get the remainder is one, the receiver assumes
error occur in the received data. Suppose the remainder is
zero, receiver assumes no error occurs in the received data. In
proposed system, CRC is used in decoding process to improve
performance of the turbo codes. The turbo decoder iterates
until CRC is verified. Once if the CRC is verified the process
stops [1] (see Fig. 4). Fig. 6. Turbo codes with MAP decoding algorithm using CRC and FC
algorithm.
A. Flip and Check Algorithm
In this paper, Flip and Check Algorithm is used to find
the exact position of the least reliable bits present in the MAP decoding algorithm in the presence of Rayleigh channel
generated codeword. The least bits is identified by using and the BPSK modulation scheme is used. The proposed CRC
state and branch metric values of the extrinsic information. and FC algorithm is the important method to improve BER
In the proposed system, Flip and check algorithm is used performance of the turbo codes (see Fig. 5).
to improve the performance and decreasing the time taken In Fig. 6 shows the BER performance of MAP decoding
by decoding process. In MAX Log MAP gives better BER algorithm over Rayleigh fading channel. The code rate is
perormance after using this algorithm in decoding process. 1/3; the number of iteration up to 8 and the block size is
Le1 represents the extrinsic information of first SISO decoder 6144 bits. MAP decoding has a poor BER performance and
and Le2 implies that second decoder output. The reliable high computational complexity so it is not suited for many
bits identification based on Eq. (10). The FC algorithm first wireless communications.
iterates minimum times without any CRC verification. The In Fig. 7 shows the BER performance of Max Log MAP
process continues until the CRC is verified. Once, the CRC decoding algorithm over Rayleigh fading channel. The code
is verified automatically stops the process. The code word rate is 1/3; the number of iteration up to 8 and the block
identification is based on presence of reliable bits in the size is 1024 bits. It has a better BER performance and
extrinsic information [1]. less computational complexity so it is well suited for many
Ae(k) = Le1(k) + Le2(k). (10) wireless communications such as LTE standard. The bit error
rate is decreased when increasing the number of iteration in
V. S IMULATION R ESULTS decoding process.
The simulated results were obtained using Matlab software. In Fig. 8 shows the comparison of BER performance for
The following results are analyzed for MAP and MAX Log MAP and Max Log MAP decoding algorithm using proposed

629
 This full-text paper was peer-reviewed and accepted to be presented at the IEEE WiSPNET 2017 conference.
Fig. 7. Turbo codes with Max Log MAP decoding algorithm using CRC and
FC algorithm.
Fig. 8. Performance comparison between the MAP and MAX-Log MAP
decoding algorithm of turbo codes using CRC and FC algorithm.
CRC and FC algorithm. After using the CRC and FC algorithm
in decoding process the bit error rate get decreased. The code
rate required for this process 1/3. From the simulation analysis,
the MAX log MAP decoding algorithm is the best choice for
LTE standard.
In Fig. 9 shows the performance of interleaver block di-
mensions by varying the length of the inter-leaver. Length of
inter-leaver is large it will randomize the errors and improve
the error performance in turbo codes. Here, the length of the
interleaver is 40 and 512 and the number of iterations required
up to 3. Before using the CRC and FC algorithm in decoding
process, even though size of the interleaver is large the bit
error rate has not decreased.
In Fig. 10 shows the performance of interleaver block
dimensions by varying the length of the inter-leaver. Length
of inter-leaver is large it will improve the error performance
in turbo codes. Here, the length of the interleaver is 40 and
512 and the number of iteration required up to 3. After using
CRC and FC algorithm in decoding process, the bit error rate
get decreased when increasing the size of the interleaver.
630
Fig. 9. Performance comparison between interleaver block dimensions before
using CRC and FC algorithm in decoding process.
Fig. 10. Performances comparison between interleaver block dimensions after
using CRC and FC algorithm in decoding process.
VI. C ONCLUSION
Thus the BER performance of LTE turbo codes is eval-
uated using CRC and FC algorithm in decoding process
over Rayleigh fading channel. The proposed algorithm for
turbo codes shows improvement in the BER performance.
When analyzing the simulated results, comparing with MAP
decoding algorithm, MAX Log MAP algorithm provides better
BER performance and it has less computational complexity. In
the proposed work, random interleaver and MAX Log MAP
decoding algorithm is used. The type of interleaver and choice
of decoding algorithm is important to improve the performance
in turbo codes.
R EFERENCES
[1] Thibaud Tonnellier, Camille Leroux, and Bertrand Le Gal, “Lowering
the error floor of turbo codes with CRC verification,” IEEE Transaction
on Wireless Communications, vol. 5, no. 4, 2016.
[2] Bernard Sklar, “Digital communication fundamentals and applications,”
Communication engineering services Tarzana, California.
[3] Claude Berrou, Alain Glavieux, and Punya Thitimajshima, “Near shan-
non limit error - correcting coding and decoding: turbo-codes,” in IEEE
International Conference on Communication, 1993, vol. 2.
[4] Cristian Anghel, Valentin Stanciu, and Constantin Paleologu, “CTC
turbo decoding architecture for LTE systems implemented on FPGA,”
in Eleventh International Conference on Networks, 2012.
[5] Hyundong Shin and Sunghwan Kim, “Turbo decoding in a Rayleigh fad-
ing channel with estimated channel state information,” IEEE Transaction
on Information Theory, 2000.
 This full-text paper was peer-reviewed and accepted to be presented at the IEEE WiSPNET 2017 conference.
[6] Stewart Crozier and Paul Guinand, “High-performance low-memory
interleaver banks for turbo-codes,” in Communications Research Centre,
Ottawa, Canada.
[7] R. Garello, P. Pierleoni, and S. Benedetto, “Computing the free dis-
tance of turbo codes and serially concatenated codes with interleavers:
algorithms and applications,” IEEE Journal on Communication, vol. 19,
no. 5, 2001.
[8] Fred Daneshgaran and Massimiliano Laddomada, “Interleaver design
for serially concatenated convolutional codes,” IEEE Transactions on
Information Theory, vol. 50, no. 6, June 2004.
[9] Fred Daneshgaran and Massimiliano Laddomada, “Reduced complexity
interleaver growth algorithm for turbo codes,” IEEE Transactions on
Wireless Communications, vol. 4, no. 3, May 2005.
[10] M. Campanella, G. Garbo, and S. Mangione, “Simple method for
limiting delay of optimized interleavers for turbo codes,” IEEE
Electronics Letter., vol. 36, no. 14, pp. 1216–1217, July 2000.
[11] Jungpil Yu and Marie-Laure Boucheret, “Interleaver design for
turbo codes from convergence analysis,” IEEE Transactions on
Communications, vol. 54, no. 4, Apr. 2006.
[12] Krishna R. Narayanan and Gordon L. Stüber, “List decoding of turbo
codes,” IEEE Transaction on Communication, vol. 46, no. 6, June 1998.
631
[13] Youssouf Ould-Cheikh-Mouhamedou, Stewart Crozier, and Ken Gracie,
“A method for lowering turbo code error flare using correction impulses
and repeated decoding,” in Communications Research Centre, Ottawa,
Canada, Apr. 3–7 2006.
[14] Joachim Hagenauer and Lutz Papke, “Iterative decoding of binary block
and convolutional codes,” IEEE Transactions on Information Theory,
vol. 42, no. 2, Mar. 1996.
[15] Andrew C. Reid, T. Aaron Gulliver, and Desmond P. Taylor,
“Convergence and errors in turbo-decoding,” IEEE Transaction on
Communication, vol. 49, no. 12, Dec. 2001.
[16] Carl Fredrik Leanderson and Carl-Erik W. Sundberg, “On list sequence
turbo decoding,” IEEE Transaction on Communication, vol. 53, no. 5,
May 2005.
[17] Jung-Fu Cheng and Havish Koorapaty, “Error detection reliability of
LTE CRC coding,” in Ericsson research, RTP, NC, USA.
[18] Lucian Andrei and Rodica Stoian, “The decision reliability of MAP,
Log-MAP, Max-Log-MAP and SOVA algorithms for turbo codes,” in
International Journal of Communications, vol. 2, no. 1, 2008.
[19] Jorge Ortín and Antonio Valdovinos, “Performance analysis of turbo
decoding algorithms in wireless OFDM systems,” IEEE Transactions
on Consumer Electronics, vol. 55, no. 3, Aug. 2009.