![International Journal of Reconfigurable and Embedded Systems (IJRES)
Vol. 12, No. 2, July 2023, pp. 174~185
ISSN: 2089-4864, DOI: 10.11591/ijres.v12.i2.pp174-185 174
Journal homepage: http://ijres.iaescore.com
Turbo encoder and decoder chip design and FPGA device
analysis for communication system
Aakanksha Devrari, Adesh Kumar
Department of Electrical and Electronics Engineering, School of Engineering, University of Petroleum and Energy Studies,
Dehradun, India
Article Info ABSTRACT
Article history:
Received Sep 2, 2022
Revised Oct 15, 2022
Accepted Jan 25, 2023
Turbo codes are error-correcting codes with performance that is close to the
Shannon theoretical limit (SHA). The motivation for using turbo codes is
that the codes are an appealing mix of a random appearance on the channel
and a physically realizable decoding structure. The communication systems
have the problem of latency, fast switching, and reliable data transfer. The
objective of the research paper is to design and turbo encoder and decoder
hardware chip and analyze its performance. Two convolutional codes are
concatenated concurrently and detached by an interleaver or permuter in the
turbo encoder. The expected data from the channel is interpreted iteratively
using the two related decoders. The soft (probabilistic) data about an
individual bit of the decoded structure is passed in each cycle from one
elementary decoder to the next, and this information is updated regularly.
The performance of the chip is also verified using the maximum a posteriori
(MAP) method in the decoder chip. The performance of field-programmable
gate array (FPGA) hardware is evaluated using hardware and timing
parameters extracted from Xilinx ISE 14.7. The parallel concatenation offers
a better global rate for the same component code performance, and reduced
delay, low hardware complexity, and higher frequency support.
Keywords:
FPGA synthesis
MAP algorithm
Turbo codes
VHDL
Wireless communication
Xilinx simulation
This is an open access article under the CC BY-SA license.
Corresponding Author:
Adesh Kumar
Department of Electrical and Electronics Engineering, School of Engineering
University of Petroleum and Energy Studies
Via Prem Nagar, Bidholi Campus Dehradun, Pin-248007, India
Email: adeshmanav@gmail.com
1. INTRODUCTION
In the present telecommunication systems, the transmission of data at high speed, voice quality, and
expansion of broadband services are the most challenging areas for improvement. Thus, the evolution of
technology from the first generation to the third, third to fourth, and now the fifth generation has experienced
a dynamic change in data transmission rate [1]. Another important aspect of any wireless communication
system [2] is its reliable data transmission means data transmission and reception should be of the same
quality. During transmission, the data is subjected to noise or distortion when transmitted via the channel.
The noise corrupts the data [3] due to which failure in data transmission occurs. Therefore, error detection
and error correction [4] play an important role in data communication. For this purpose, different channel
coding techniques are used. Some of these coding methods are low density parity check (LDPC),
convolution, polar, and turbo codes [5]. The performance of these codes can be analyzed by their
performance in expressions of bit error rate, complexity, functionality, and hardware device utilization. In the
late 1940s, a researcher named Shannon gave a concept describing the theoretical limit [6] of maximum
capacity for data transmission in any communication channel. This channel capacity limit has now become](https://image.slidesharecdn.com/0320622-39762-1-pb-230720025712-a2b43fce/85/Turbo-encoder-and-decoder-chip-design-and-FPGA-device-analysis-for-communication-system-1-320.jpg)
![Int J Reconfigurable & Embedded Syst ISSN: 2089-4864
Turbo encoder and decoder chip design and FPGA device analysis for … (Aakanksha Devrari)
175
the goal for any communication system. To better understand the need for this theoretical limit, the modern
mobile industry can be considered. As the number of users is emergent enormously day by day the need and
the complexity of the application are also increasing? For example, earlier mobile phones were only used for
making calls or text messages but now we can access multiple features like sending pictures, and videos, and
using the internet. This technological advancement demands a higher data rate.
As the number of consumers are emergent to access the existing channel different multiplexing
techniques are used such as frequency division multiple access (FDMA), time division multiple access
(TDMA), and code division multiple access (CDMA). However, these techniques do not increase the rate of
data transmission. To increase the data rate one way is to increase the bandwidth of the passage but this
reduces the number of available channels thus, this method does not solve the problem. Another way to
increase the data rate is by improving the coding scheme so that the existing channel can be utilized to its full
strength. This allows more channels to accommodate a particular range of bandwidth. As shown by Berrou
[2] in his research work where a new category of code was introduced that performs close to Shannon’s limit
[6]. The proposed work stated turbo codes based on an iterative decoding scheme. Turbo encoder consists of
two convolution encoders connected in equivalent and detached by an interleaver. Moreover, the receiver
consists of the serial concatenation of convolution decoders. These decoders use the data from the previous
decoder for the error-controlling process. The whole iteration process can be repeatedly run for one received
codeword until further improvement. For the iterative decoding mechanism, two algorithms are present: the
soft output viterbi algorithm (SOVA) and the maximum a posteriori probability (MAP) algorithm [7]. Earlier
application-specific integrated circuits (ASICs) were considered the standard structure for implementing
wireless transmitters and receivers. For multiple wireless standards, devices require separate ASICs for each
wireless standard which makes the circuit more complex, costly, and power-consuming. The wireless
standards have limited use of fixed ASIC and demand a new flexible architecture. The flexible architectures
share hardware for common wireless standard features. Therefore, constraints like reduced chip area,
complexity, and low power consumption can be optimized while using flexible architecture.
Various reconfigurable architectures, such as field-programmable gate array (FPGA) [8] are now
available with user-programmable logic blocks that may be altered based on the requirements or applications.
These architectures are attractive solutions to get the desired functionality with relatively high flexibility. The
error correction method [9] is incorporated into a wireless communication system for reliable data
transmission in channels subjective to noise. The encoding of the input data structure is done by the
transmitter based on a particular encoding scheme. This encoding adds some redundant bits/data in the data
sequence and at the receiver end, these encoded bits are separated for reliable detection of the data sequence.
In any wireless channel whenever data bits are transmitted, they are subjective to the noise that causes the
error. Forward error correction codes can be used to detect and rectify this problem. This work is
accomplished by encoders such as convolution or block codes, which add redundant bits to the data
sequence. If an error occurs during transmission, the decoder detects it and corrects it at the receiver.
Turbo codes were presented in 1993 as a new type of error correction code. These codes are made
up of two or more recursive systematic convolutional encoders [10] coupled in equivalent and detached by an
interleaver. Turbo codes perform better in relation to bit error rate (BER) than convolution codes because of
their concatenation structure and iterative decoding method. Turbo codes are used in existing wireless
protocols such as long term evolution (LTE), universal mobile telecommunication system (UMTS), and high-
speed downlink packet access (HSDPA). A binary turbo encoder with two recursive convolution encoders is
shown in Figure 1. Two parity bits are generated for each message bit given to the turbo encoder. The
production of the turbo encoder is the interleaved version of input bits. An interleaver is employed because
consecutively transmitted bits can be influenced by faulty bits. The production of the turbo encoder is
comprised of parity bits produced by recursive convolution encoder-1 (RSE-1) and recursive convolution
encoder-2 (RSE-2). A systematic frame and a parity frame are generated by two convolution encoders. The
feedforward polynomial (1+m+m3
) is generated by RSE-1 and feedback generator polynomial (1+m2
+m3
) is
generated by RSE-2. The value of the code rate for each encoder is R=1/3, and the length of the sequence is
3N bits. A state diagram with (n=8) is considered for the turbo encoder as revealed in Figure 2. Initially the
state of the turbo encoder is (S0=‘0’) and S(n) ϵ {0, 1… 7} are the subsequent states for S(n) ϵ {0, 1} message
bit. The states are S0(000), S1(001), S2(010), S3(011), S4(100), S5(101), S6(110), and S7(111) corresponding to
binary message inputs (di=0, di=1).
The interleaver is the main block in the turbo encoding and decoding process. Due to the
interleaving process burst errors in consecutive bits are spread over the transmitted sequence. Thus, bit error
rate performance is improved as the bits are reordered in some pseudo-random manner. Another process is
puncturing, which increases the code rate by removing the coded parity bits of the sequence thus, making the
channel utilization efficient. The interleaver in the turbo encoder structure generates a long data block by
reshuffling the data and the interleaver in the decoder structure corrects the errors by bypassing the data to
the first decoder. After this process, the interleaver further reshuffles the firstly decoded data and once](https://image.slidesharecdn.com/0320622-39762-1-pb-230720025712-a2b43fce/85/Turbo-encoder-and-decoder-chip-design-and-FPGA-device-analysis-for-communication-system-2-320.jpg)
![ ISSN: 2089-4864
Int J Reconfigurable & Embedded Syst, Vol. 12, No. 2, July 2023: 174-185
176
shuffling is done passes the data to the second decoder for remaining error corrections. The same way
process is repeated.
The organization of the manuscript is as follows. Section 2 offerings the related work with the
identification of the gaps, problem statement, and originality. Section 3 offerings the proposed method.
Section 4 details the methodology and implementation strategy followed by parallel decoding and
interleaving design architecture. Section 5 offerings the results and deliberations part with the possible
simulation test inputs and analysis for the FPGA parameters followed by conclusions in section 6.
Figure 1. Turbo encoder [11]
Figure 2. Trellis diagram configuration [11]
2. RELATED WORK
Turbo codes are used in 4G mobile infrastructures as well as in deep-space satellite communications
and other applications where inventors want to achieve dependable data transmission over bandwidth-or
latency-inhibited transmission links in the existence of data-corrupting noise. The hardware application of a
serial turbo decoder architecture [12] was done on FPGA for LTE and worldwide interoperability for
microwave access (Wi-MAX) communication system. An interleaver memory (ILM) was used for LTE
systems that utilized comparators and subtractors instead of multipliers and dividers. This unique interleaver
overcomes the need for memory utilization as compared to the parallel architecture approach. The algorithm
was proposed for the scheme of the interleaver [13] for turbo decoder architecture. The error correction
performance, as well as latency, was improved in comparison to the sequential decoder by the integration of
the algorithm. The algorithm involves the parallel architecture structure which is highly efficient in relations
of area, and BER thus, improving the throughput respectively.
Very high-speed integrated circuit hardware description language (VHDL) was used for the
application of different error control coding [14] encoders and decoders on FPGA. The error control codes is](https://image.slidesharecdn.com/0320622-39762-1-pb-230720025712-a2b43fce/85/Turbo-encoder-and-decoder-chip-design-and-FPGA-device-analysis-for-communication-system-3-320.jpg)
![Int J Reconfigurable & Embedded Syst ISSN: 2089-4864
Turbo encoder and decoder chip design and FPGA device analysis for … (Aakanksha Devrari)
177
an important area of discussion in telecommunication scenario since it provides efficient ways of error
correction and error detection. The analysis of different channel coding schemes [15] was proposed for the
fifth-generation (5G) mobile system based on binary phase-shift keying (BPSK) modulation over Additive
white Gaussian noise (AWGN) channels such as polar, turbo, and low-density parity-check (LDPC). In terms
of computational complexity, polar codes proved to be the best of all. A turbo decoder architecture was
presented based on the log MAP algorithm [16] using a parallel processor to meet the requirements of long-
term evolution communication standards. It is observed that this structure gives better performance in
relation to BER. The synthesis results show the device utilization such as the count values of slices and 4
input lookup tables (LUTs) is less using this algorithm, making the design cost-effective. Based on the study
and discussion on the forward error correction schemes [17] such as turbo codes and conventional coding, it
is observed that turbo codes depict the best theoretical bound in terms of efficiency and channel capacity. In
recent times these codes are preferred in several applications such as CDMA-2000 and 3G cellular networks
such as UMTS.
The log-MAP algorithm was used for turbo decoding [18] in the Wi-Max communication system.
The lookup table (LUT) based log-MAP algorithm was proposed for turbo decoder structure in LTE wireless
communication system [19] that reduces the logic resource utilization by 15% in comparison to the
conventional design approach. The implementation of a concurrent turbo decoder [20] with an inverse
address originator scheme was proposed for future broadcasting communication systems. This address-
generation scheme reduces the iteration processing time which further reduces latency. Xilinx integrated
logic analyzer (ILA) tool is used to check the validity of the turbo decoder, block error rate (BLER), and
latency. A new design approach for the realization of a turbo decoder on [21] FPGA with a reverse address
generator to reduce the complexity and latency.
To reduce memory issues a contention-free interleaver is designed and a clock gating technique is
used to improve power dissipation. The turbo decoder base was proposed to get high throughput. This was
achieved using a processor in which the algorithm has accelerator units and highly controlled software. As
the need for configurable memory interfacing is required memory interfacing with a switch fuse unit is used
instead of fixed memory interfacing. The parallel decoder [22] based on configurable interleaved networking
was used for a long-term evolution standard. A multi-parallelism interleaver approach [23] was designed to
reduce computational complexity. The performance loss due to parallel structure is compensated using path
metrics [24]. The proposed design is a competitive solution to recent work in terms of high reliability and
efficiency.
The related work presents the turbo decoder design and FPGA implementation with different
aspects. They have implemented the LTE turbo decoder with the Max log-MAP technique on Virtex-6
XC6VLX75T with a clock frequency of 270 MHz, the number of slices 4108, and LUTs 6310. The work is
also done on the VHDL-based design to implement a turbo decoder with a log-MAP technique that follows
the decoding iterations but not analyzed FPGA resource utilization. The performance of the turbo codec is
also estimated with the log-MAP algorithm on central processing unit (CPU) and FPGA and further
utilization of the FPGA for LTE. The work is not done on the latest FPGA that provides fast switching,
minimum FPGA hardware parameters utilization, memory, and delay. The problem declaration of the study
is to analyze the performance of the hardware chip design of the turbo encoder and decoder based on simple
architecture by the integration of the log-MAP algorithm in the same chip, perform the Xilinx simulation for
the data verification, and compare the chip design performance on latest supporting FPGA with hardware and
timing parameters. The originality of the work is that the encoder and decoder design supports greater
frequency with optimal timing and hardware FPGA resources.
3. PROPOSED ALGORITHM
The decoder structure is a parallel concatenated arrangement of convolution codes that takes the
independent set of parity bits. After a systematic bit sequence is formed this process is known as the iterative
decoding method [25]. The iterative decoding mechanism and the configuration of the turbo decoder are
shown with the help of Figure 3 and Figure 4. The information or the message bits are processed between
soft input and soft-output (SISO)-1 to SISO-2. Extrinsic information generated by SISO-1 is de-interleaved
and fed as input to the next SISO-2. This building block calculates the forward matrix value (α), backward
matrix value (β), and the output SISO matrix [5]. To calculate α and β values the Bahl, Cocke, Jelinek, and
Raviv (BCJR) algorithm is applied. The SISO decoders perform the turbo decoding process and are referred
to as a log-likelihood ratio (LLR) [26]. For the encoded sequence, di = [d1, d2…dN], and generated codeword
against each sequence di = [d1, d2…dN], the LLR is articulated as (1).](https://image.slidesharecdn.com/0320622-39762-1-pb-230720025712-a2b43fce/85/Turbo-encoder-and-decoder-chip-design-and-FPGA-device-analysis-for-communication-system-4-320.jpg)
![ ISSN: 2089-4864
Int J Reconfigurable & Embedded Syst, Vol. 12, No. 2, July 2023: 174-185
178
Figure 3. Turbo decoding [5]
Figure 4. Iterative turbo decoding [26]
𝐿(𝑑𝑖) = ln
𝑃(𝑑𝑖=1)
𝑃(𝑑𝑖=0)
(1)
For the decoded sequence h=[h1, h2…hN],
𝐿(𝑑𝑖|ℎ) = ln
𝑃(𝑑𝑖=1|ℎ)
𝑃(𝑑𝑖=0|ℎ)
(2)
Bahl et al. [1] discussed the MAP algorithm in 1974, which was applied to breed the probabilities of
individual bits and descend the extrinsic information. The LLR of the kth
symbol is given as (3).
𝐿(𝑑𝑘|ℎ) = 𝑙𝑛
∑ 𝛼𝑘−1
′
𝑑𝑘=1 (𝑠𝑘−1)𝛾𝑘
′(𝑠𝑘−1,𝑠𝑘)𝛽𝑘
′ (𝑠𝑘)
∑ 𝛼𝑘−1
′
𝑑𝑘=0 (𝑠𝑘−1)𝛾𝑘
′(𝑠𝑘−1,𝑠𝑘)𝛽𝑘
′ (𝑠𝑘)
(3)
In the (3):
𝛼𝑘
′
=Forward state matric
𝛽𝑘
′
=Backward state matric
𝛾𝑘
′
=Branch matric
𝑠𝑘=Trellis diagram states at the time ‘k’
The MAP algorithm [26] comprises addition processes and multiplication processes. For a large
classification, the logarithm and estimate are applied, and the (3) for Max log-MAP is specified as (4).
ln(∑ 𝑒𝑔𝑖) ≈ 𝑚𝑎𝑥𝑖𝑚𝑢𝑚 (𝑔𝑖) (4)
Then the algorithm is streamlined to the (5).
𝐿(𝑑𝑘|ℎ) = 𝑚𝑎𝑥𝑑𝑘=1(𝛼𝑘−1 𝛾𝑘(𝑠𝑘−1, 𝑠𝑘) 𝛽𝑘 (𝑠𝑘)) − 𝑚𝑎𝑥𝑑𝑘=0(𝛼𝑘−1 𝛾𝑘(𝑠𝑘−1, 𝑠𝑘) 𝛽𝑘 (𝑠𝑘)) (5)
The expression has addition operations and association operations. The state matrix is similarly condensed
and written as (6) and (7).
𝛼𝑘 (𝑠𝑘) = ln( 𝛼𝑘
′
(𝑠𝑘)) = 𝑚𝑎𝑥𝑖𝑚𝑢𝑚 (𝛼𝑘−1(𝑠𝑘−1) + 𝛾𝑘−1(𝑠𝑘−1)) (6)](https://image.slidesharecdn.com/0320622-39762-1-pb-230720025712-a2b43fce/85/Turbo-encoder-and-decoder-chip-design-and-FPGA-device-analysis-for-communication-system-5-320.jpg)
![ ISSN: 2089-4864
Int J Reconfigurable & Embedded Syst, Vol. 12, No. 2, July 2023: 174-185
180
is used. For the selection of SISO and first input first output (FIFO) modules switch matrices are used.
Depending upon the priority, each FIFO module is attached to its current SISO module for input-output
buffer synchronization. Since, the LLR, branch matrices, forward matrices, and backward matrices
parameters are sequential. The throughput of the turbo decoder can be maximized using input and output
buffers [27].
The design methodology of the work is as follows. The chip functionality follows the steps of the
chip design, and simulation modeling. The VHDL programming is used to design the turbo encoder and
decoder chip that accepts the 32-bit input stream. The waveform simulation is carried out for both the
integrated decoder module in which the test cases are verified. The real-time verification is done on the
Virtex-5 FPGA board to check the feasibility of the design. The method for the same is given in Figure 6.
Figure 6. Method
5. RESULTS AND DISCUSSIONS
In Xilinx ISE 14.7, the behavioral modeling-based model is used to develop the turbo encoder and
decoder. The encoder and decoder integrated chip register transfer level (RTL) view are shown in Figure 7.
Table 1 lists the specifics of RTL pins. Figure 8 depicts the waveform in Xilinx ISIM for the decoder for test-
1 and test-2 in binary without error in reception. Figure 9 depicts the waveform in Xilinx ISIM for the
decoder for test-1 and test-2 in hexadecimal without error in reception. Figure 10 depicts the waveform in
Xilinx ISIM for the decoder for test-3 and test-4 in binary with an error in reception. Figure 11 depicts the
waveform in Xilinx ISIM for the decoder for test-3 and test-4 in hexadecimal with an error in reception. The
major inputs and outputs of the decoder's main stimuli are displayed on the simulation test bench.
Table 1. Pin description of encoder module and decoder module RTL chip
Pins Direction Description
Clk Input The clock signal has a 50% duty cycle and is an input clock signal that can
provide a rising edge-based clock signal.
Reset Input To maintain the contents 0 for the encoder and decoder, reset is the chip
design's default input.
Turbo_tx_data_in_encoder [31:0] Input It gives the turbo encoder provided at the transmitter section of the design
the 32-bit input message.
Turbo_rx_data_out_decoder [31:0] Output The 32-bit turbo-encoded output can be treated as an AWGN channel's
input.
Turbo_enc_data_inout [31:0] Inout It is the 32-bit output of the AWGN communication channel and feedback
to the turbo decoder.
Error Output It is the indicator output that a correct or incorrect message is received in
the decoder section.](https://image.slidesharecdn.com/0320622-39762-1-pb-230720025712-a2b43fce/85/Turbo-encoder-and-decoder-chip-design-and-FPGA-device-analysis-for-communication-system-7-320.jpg)
![ ISSN: 2089-4864
Int J Reconfigurable & Embedded Syst, Vol. 12, No. 2, July 2023: 174-185
182
Test case (turbo encoder/ decoder):
− Apply Clk=rising edge clock signal, reset=‘0’, turbo_tx_data_in_encoder [31:0]=“0000 0000 0000 0000
1100 1100 0000 0000” (binary)=0000CC00 (hexadecimal), then turbo_enc_data_inout [31:0]=“0000
0000 0000 0000 1100 1100 0000 0000” (binary)=0000CC00 (hexadecimal), and
turbo_rx_data_out_decoder [31:0]=“0000 0000 0000 0000 1100 1100 0000 0000” (binary)=0000CC00
(hexadecimal), with error=‘0’
− Apply Clk=rising edge clock signal, reset=‘0’, turbo_tx_data_in_encoder [31:0]=“0000 0000 0000 0000
1010 1111 0000 0000” (binary)=0000AF00 (hexadecimal), then turbo_enc_data_inout [31:0]=“0000
0000 0000 0000 1010 1111 0000 0000” (binary)=0000AF00 (hexadecimal), and
turbo_rx_data_out_decoder [31:0]=“0000 0000 0000 0000 1010 1111 0000 0000” (binary)=0000AF00
(hexadecimal) with error = ‘0’
− Apply Clk=rising edge clock signal, reset=‘0’, turbo_tx_data_in_encoder [31:0]=“0000 0000 0000 0000
1111 0010 0000 0000” (binary)=0000F200 (hexadecimal), then turbo_enc_data_inout [31:0]=“0000
0000 0000 0000 1111 0001 0000 0000” (binary)=0000F100 (hexadecimal), and
turbo_rx_data_out_decoder [31:0]=“0000 0000 0000 0000 1111 0010 0000 0000” (binary)=0000F200
(hexadecimal) with error=‘1’
− Apply Clk=rising edge clock signal, reset=‘0’, turbo_tx_data_in_encoder [31:0]=“0000 0000 0000 0000
0011 1111 0000 0000” (binary)=00003F00 (hexadecimal), then turbo_enc_data_inout [31:0]=“0000
0000 0000 0000 0010 1111 0000 0000” (binary)=00002F00 (hexadecimal), and
turbo_rx_data_out_decoder [31:0]=“0000 0000 0000 0000 0011 1111 0000 0000” (binary)=00003F00
(hexadecimal) with error=‘1’.
Table 2 summarizes the hardware parameters of the turbo encoder and decoder hardware with MAP
algorithm in Xilinx ISE 14.7 software for various parameters targeting the Virtex-5 FPGA. Slices, flip-flops,
LUTs, and input/output block (IOBs) are the hardware parameters [28]. Table 3 shows the results of timing
simulations for the turbo encoder and decoder chip hardware for time-concerned parameters such as
minimum period (ns), minimum and maximum duration before and after the clock signal (ns), and maximum
supporting frequency [29].
Table 2. Summary of FPGA hardware usage
Hardware Turbo codes Turbo codes with Max log-MAP
Encoder Decoder Encoder Decoder
Slices 75 150 67 140
Flip-flops 87 103 77 80
LUTs 120 125 103 107
IOBs 64 96 64 96
Global Clocks (GCLKs) 1 1 1 1
Table 3. Parameters connected to timing information
Timing parameter Turbo codes Turbo codes with Max log-MAP
Encoder Decoder Encoder Decoder
Frequency (MHz) 314.0 320.0 332.00 355.0
Minimum period (ns) 1.710 1.420 1.6120 1.390
Time (minimum) before clock (ns) 2.109 2.917 1.964 2.851
Time (maximum) after clock (ns) 4.320 5.325 4.124 4.109
Combinational path delay (ns) 8.139 9.662 7.700 8.350
Speed grade -5 -5 -5 -5
Figure 12 presents the FPGA hardware utilization graph for the turbo encoder and decoder with a
simple and integrated Max log-MAP algorithm chip. Figure 13 presents the time-related parameters of FPGA
hardware. The combinational path delay and related time parameters are reduced by the integration of the
Max log-MAP algorithm in the turbo encoder and decoder hardware due to parallel execution.
The major motivation for parallel implementation is to efficiently execute code, as it saves time and
allows programmers to be executed in a shorter clock time. There is significantly less disruption for workers
and equipment because multiple processes can run simultaneously. The encoder and decoder both utilize less
hardware resources in comparison to the turbo encoder and decoder due to the parallel execution of the block.
The designed hardware chip reduces the hardware utilization in the FPGA synthesis.](https://image.slidesharecdn.com/0320622-39762-1-pb-230720025712-a2b43fce/85/Turbo-encoder-and-decoder-chip-design-and-FPGA-device-analysis-for-communication-system-9-320.jpg)
![ ISSN: 2089-4864
Int J Reconfigurable & Embedded Syst, Vol. 12, No. 2, July 2023: 174-185
184
REFERENCES
[1] L. Bahl, J. Cocke, F. Jelinek, and J. Raviv, “Optimal decoding of linear codes for minimizing symbol error rate,” IEEE
Transactions on Information Theory, vol. 20, no. 2, pp. 284–287, Mar. 1974, doi: 10.1109/TIT.1974.1055186.
[2] C. Berrou and A. Glavieux, “Near optimum error correcting coding and decoding: turbo-codes,” IEEE Transactions on
Communications, vol. 44, no. 10, pp. 1261–1271, 1996, doi: 10.1109/26.539767.
[3] R. Mishra, P. Kuchhal, and A. Kumar, “Effect of height of the substrate and width of the patch on the performance characteristics
of microstrip antenna,” International Journal of Electrical and Computer Engineering (IJECE), vol. 5, no. 6, pp. 1441–1445, Dec.
2015, doi: 10.11591/ijece.v5i6.pp1441-1445.
[4] K. H. Kaittan and S. J. Mohammed, “Implementing and designing a secure information system based on the dsss gold sequence
using matlab,” Bulletin of Electrical Engineering and Informatics, vol. 10, no. 3, pp. 1455–1463, Jun. 2021, doi:
10.11591/eei.v10i3.2446.
[5] M. J. Adamu, L. Qiang, R. S. Zakariyya, C. O. Nyatega, H. B. Kawuwa, and A. Younis, “An efficient turbo decoding and
frequency domain turbo equalization for LTE based narrowband internet of things (NB-IoT) systems,” Sensors, vol. 21, no. 16,
Aug. 2021, doi: 10.3390/s21165351.
[6] C. E. Shannon, “A mathematical theory of communication,” Bell System Technical Journal, vol. 27, no. 3, pp. 379–423, 1948,
doi: 10.1002/j.1538-7305.1948.tb01338.x.
[7] L. Hang, G. Zhen, Z. Ming, and W. Jing, “Partial iterative decode of turbo codes for on-board processing satellite platform,”
China Communications, vol. 12, no. 11, pp. 1–8, Nov. 2015, doi: 10.1109/CC.2015.7366233.
[8] L. Fanari, E. Iradier, I. Bilbao, R. Cabrera, J. Montalban, and P. Angueira, “Comparison between different channel coding
techniques for IEEE 802.11be within factory automation scenarios,” Sensors, vol. 21, no. 21, Oct. 2021, doi: 10.3390/s21217209.
[9] P. Salija, B. Yamuna, T. R. Padmanabhan, and D. Mishra, “Generic direct approach for decoding turbo codes using probability
density based reliability model,” Journal of Communications Technology and Electronics, vol. 66, no. 2, pp. 175–183, Feb. 2021,
doi: 10.1134/S1064226921020133.
[10] A. Devrari, A. Kumar, A. Kumar, and S. Singh, “Design and FPGA implementation of DSSS for near-far effect in MANET,” in
Proceeding of International Conference on Intelligent Communication, Control and Devices . Advances in Intelligent Systems and
Computing, Singapore: Springer, 2017, pp. 425–434.
[11] L. F. Gonzalez-Perez, L. C. Yllescas-Calderon, and R. Parra-Michel, “Parallel and configurable turbo decoder implementation for
3GPP-LTE,” in 2013 International Conference on Reconfigurable Computing and FPGAs (ReConFig), Dec. 2013, pp. 1–6, doi:
10.1109/ReConFig.2013.6732316.
[12] J. P. Woodard and L. Hanzo, “Comparative study of turbo decoding techniques: an overview,” IEEE Transactions on Vehicular
Technology, vol. 49, no. 6, pp. 2208–2233, 2000, doi: 10.1109/25.901892.
[13] R. Dobkin, M. Peleg, and R. Ginosar, “Parallel VLSI architecture for MAP turbo decoder,” in The 13th IEEE International
Symposium on Personal, Indoor and Mobile Radio Communications, 2002, pp. 384–388, doi: 10.1109/PIMRC.2002.1046727.
[14] J. M. Gilbert, C. Robbins, and W. Sheikh, “FPGA implementation of error control codes in VHDL: an undergraduate research
project,” Computer Applications in Engineering Education, vol. 27, no. 5, pp. 1073–1086, Sep. 2019, doi: 10.1002/cae.22137.
[15] Z. R. M. Hajiyat, A. Sali, M. Mokhtar, and F. Hashim, “Channel coding scheme for 5G mobile communication system for short-
length message transmission,” Wireless Personal Communications, vol. 106, no. 2, pp. 377–400, May 2019, doi: 10.1007/s11277-
019-06167-7.
[16] S. Kalyani, V. Raj, and K. Giridhar, “Narrowband interference mitigation in turbo-coded OFDM systems,” in 2007 IEEE
International Conference on Communications, Jun. 2007, pp. 1059–1064, doi: 10.1109/ICC.2007.180.
[17] J. D. Kene and K. D. Kulat, “Soft output decoding algorithm for turbo codes implementation in mobile Wi-Max environment,”
Procedia Technology, vol. 6, pp. 666–673, 2012, doi: 10.1016/j.protcy.2012.10.080.
[18] J. Li, X. Wang, J. He, C. Su, and L. Shan, “Turbo decoder design based on an LUT-normalized Log-MAP algorithm,” Entropy,
vol. 21, no. 8, Aug. 2019, doi: 10.3390/e21080814.
[19] H. Luo et al., “Low latency parallel turbo decoding implementation for future terrestrial broadcasting systems,” IEEE
Transactions on Broadcasting, vol. 64, no. 1, pp. 96–104, Mar. 2018, doi: 10.1109/TBC.2017.2704425.
[20] V. Akshaya, K. N. Sreehari, and A. Chalil, “VLSI implementation of turbo coder for LTE using verilog HDL,” in 2020 Fourth
International Conference on Computing Methodologies and Communication (ICCMC), Mar. 2020, pp. 275–279, doi:
10.1109/ICCMC48092.2020.ICCMC-00051.
[21] P. Salmela, H. Sorokin, and J. Takala, “A programmable max-log-MAP turbo decoder implementation,” VLSI Design, vol. 2008,
pp. 1–17, Dec. 2008, doi: 10.1155/2008/319095.
[22] Z. Yan, G. He, W. He, S. Wang, and Z. Mao, “High performance parallel turbo decoder with configurable interleaving network
for LTE application,” Integration, vol. 52, pp. 77–90, Jan. 2016, doi: 10.1016/j.vlsi.2015.05.003.
[23] G. Masera, G. Piccinini, M. R. Roch, and M. Zamboni, “VLSI architectures for turbo codes,” IEEE Transactions on Very Large
Scale Integration (VLSI) Systems, vol. 7, no. 3, pp. 369–379, Sep. 1999, doi: 10.1109/92.784098.
[24] C. Berrou, “The invention of Turbo codes: the ten-year-old Turbo codes are entering into service,” IEEE Communications
Magazine, vol. 61, no. 2, pp. 6–13, Feb. 2023, doi: 10.1109/MCOM.2023.10048673.
[25] J. Nadal and A. Baghdadi, “Parallel and flexible 5G LDPC decoder architecture targeting FPGA,” IEEE Transactions on Very
Large Scale Integration (VLSI) Systems, vol. 29, no. 6, pp. 1141–1151, Jun. 2021, doi: 10.1109/TVLSI.2021.3072866.
[26] A. A. Hamad, M. T. Gatte, and L. A. Abdul-Rahaim, “Efficient systematic turbo polar decoding based on optimized scaling factor
and early termination mechanism,” International Journal of Electrical and Computer Engineering (IJECE), vol. 13, no. 1, pp.
629–637, Feb. 2023, doi: 10.11591/ijece.v13i1.pp629-637.
[27] A. Kumar, P. Sharma, M. K. Gupta, and R. Kumar, “Machine learning-based resource utilization and pre-estimation for network
on chip (NoC) communication,” Wireless Personal Communications, vol. 102, no. 3, pp. 2211–2231, Oct. 2018, doi:
10.1007/s11277-018-5376-3.
[28] A. Kumar, G. Verma, and M. K. Gupta, “FM receiver design using programmable PLL,” Wireless Personal Communications, vol.
97, no. 1, pp. 773–787, Nov. 2017, doi: 10.1007/s11277-017-4536-1.
[29] A. Kumar, A. Kumar, and A. Devrari, “Hardware chip performance analysis of different FFT architecture,” International Journal
of Electronics, vol. 108, no. 7, pp. 1124–1140, Jul. 2021, doi: 10.1080/00207217.2020.1819441.](https://image.slidesharecdn.com/0320622-39762-1-pb-230720025712-a2b43fce/85/Turbo-encoder-and-decoder-chip-design-and-FPGA-device-analysis-for-communication-system-11-320.jpg)
Turbo encoder and decoder chip design and FPGA device analysis for communication system
- 1. International Journal of Reconfigurable and Embedded Systems (IJRES) Vol. 12, No. 2, July 2023, pp. 174~185 ISSN: 2089-4864, DOI: 10.11591/ijres.v12.i2.pp174-185 174 Journal homepage: http://ijres.iaescore.com Turbo encoder and decoder chip design and FPGA device analysis for communication system Aakanksha Devrari, Adesh Kumar Department of Electrical and Electronics Engineering, School of Engineering, University of Petroleum and Energy Studies, Dehradun, India Article Info ABSTRACT Article history: Received Sep 2, 2022 Revised Oct 15, 2022 Accepted Jan 25, 2023 Turbo codes are error-correcting codes with performance that is close to the Shannon theoretical limit (SHA). The motivation for using turbo codes is that the codes are an appealing mix of a random appearance on the channel and a physically realizable decoding structure. The communication systems have the problem of latency, fast switching, and reliable data transfer. The objective of the research paper is to design and turbo encoder and decoder hardware chip and analyze its performance. Two convolutional codes are concatenated concurrently and detached by an interleaver or permuter in the turbo encoder. The expected data from the channel is interpreted iteratively using the two related decoders. The soft (probabilistic) data about an individual bit of the decoded structure is passed in each cycle from one elementary decoder to the next, and this information is updated regularly. The performance of the chip is also verified using the maximum a posteriori (MAP) method in the decoder chip. The performance of field-programmable gate array (FPGA) hardware is evaluated using hardware and timing parameters extracted from Xilinx ISE 14.7. The parallel concatenation offers a better global rate for the same component code performance, and reduced delay, low hardware complexity, and higher frequency support. Keywords: FPGA synthesis MAP algorithm Turbo codes VHDL Wireless communication Xilinx simulation This is an open access article under the CC BY-SA license. Corresponding Author: Adesh Kumar Department of Electrical and Electronics Engineering, School of Engineering University of Petroleum and Energy Studies Via Prem Nagar, Bidholi Campus Dehradun, Pin-248007, India Email: [email protected] 1. INTRODUCTION In the present telecommunication systems, the transmission of data at high speed, voice quality, and expansion of broadband services are the most challenging areas for improvement. Thus, the evolution of technology from the first generation to the third, third to fourth, and now the fifth generation has experienced a dynamic change in data transmission rate [1]. Another important aspect of any wireless communication system [2] is its reliable data transmission means data transmission and reception should be of the same quality. During transmission, the data is subjected to noise or distortion when transmitted via the channel. The noise corrupts the data [3] due to which failure in data transmission occurs. Therefore, error detection and error correction [4] play an important role in data communication. For this purpose, different channel coding techniques are used. Some of these coding methods are low density parity check (LDPC), convolution, polar, and turbo codes [5]. The performance of these codes can be analyzed by their performance in expressions of bit error rate, complexity, functionality, and hardware device utilization. In the late 1940s, a researcher named Shannon gave a concept describing the theoretical limit [6] of maximum capacity for data transmission in any communication channel. This channel capacity limit has now become
- 2. Int J Reconfigurable & Embedded Syst ISSN: 2089-4864 Turbo encoder and decoder chip design and FPGA device analysis for … (Aakanksha Devrari) 175 the goal for any communication system. To better understand the need for this theoretical limit, the modern mobile industry can be considered. As the number of users is emergent enormously day by day the need and the complexity of the application are also increasing? For example, earlier mobile phones were only used for making calls or text messages but now we can access multiple features like sending pictures, and videos, and using the internet. This technological advancement demands a higher data rate. As the number of consumers are emergent to access the existing channel different multiplexing techniques are used such as frequency division multiple access (FDMA), time division multiple access (TDMA), and code division multiple access (CDMA). However, these techniques do not increase the rate of data transmission. To increase the data rate one way is to increase the bandwidth of the passage but this reduces the number of available channels thus, this method does not solve the problem. Another way to increase the data rate is by improving the coding scheme so that the existing channel can be utilized to its full strength. This allows more channels to accommodate a particular range of bandwidth. As shown by Berrou [2] in his research work where a new category of code was introduced that performs close to Shannon’s limit [6]. The proposed work stated turbo codes based on an iterative decoding scheme. Turbo encoder consists of two convolution encoders connected in equivalent and detached by an interleaver. Moreover, the receiver consists of the serial concatenation of convolution decoders. These decoders use the data from the previous decoder for the error-controlling process. The whole iteration process can be repeatedly run for one received codeword until further improvement. For the iterative decoding mechanism, two algorithms are present: the soft output viterbi algorithm (SOVA) and the maximum a posteriori probability (MAP) algorithm [7]. Earlier application-specific integrated circuits (ASICs) were considered the standard structure for implementing wireless transmitters and receivers. For multiple wireless standards, devices require separate ASICs for each wireless standard which makes the circuit more complex, costly, and power-consuming. The wireless standards have limited use of fixed ASIC and demand a new flexible architecture. The flexible architectures share hardware for common wireless standard features. Therefore, constraints like reduced chip area, complexity, and low power consumption can be optimized while using flexible architecture. Various reconfigurable architectures, such as field-programmable gate array (FPGA) [8] are now available with user-programmable logic blocks that may be altered based on the requirements or applications. These architectures are attractive solutions to get the desired functionality with relatively high flexibility. The error correction method [9] is incorporated into a wireless communication system for reliable data transmission in channels subjective to noise. The encoding of the input data structure is done by the transmitter based on a particular encoding scheme. This encoding adds some redundant bits/data in the data sequence and at the receiver end, these encoded bits are separated for reliable detection of the data sequence. In any wireless channel whenever data bits are transmitted, they are subjective to the noise that causes the error. Forward error correction codes can be used to detect and rectify this problem. This work is accomplished by encoders such as convolution or block codes, which add redundant bits to the data sequence. If an error occurs during transmission, the decoder detects it and corrects it at the receiver. Turbo codes were presented in 1993 as a new type of error correction code. These codes are made up of two or more recursive systematic convolutional encoders [10] coupled in equivalent and detached by an interleaver. Turbo codes perform better in relation to bit error rate (BER) than convolution codes because of their concatenation structure and iterative decoding method. Turbo codes are used in existing wireless protocols such as long term evolution (LTE), universal mobile telecommunication system (UMTS), and high- speed downlink packet access (HSDPA). A binary turbo encoder with two recursive convolution encoders is shown in Figure 1. Two parity bits are generated for each message bit given to the turbo encoder. The production of the turbo encoder is the interleaved version of input bits. An interleaver is employed because consecutively transmitted bits can be influenced by faulty bits. The production of the turbo encoder is comprised of parity bits produced by recursive convolution encoder-1 (RSE-1) and recursive convolution encoder-2 (RSE-2). A systematic frame and a parity frame are generated by two convolution encoders. The feedforward polynomial (1+m+m3 ) is generated by RSE-1 and feedback generator polynomial (1+m2 +m3 ) is generated by RSE-2. The value of the code rate for each encoder is R=1/3, and the length of the sequence is 3N bits. A state diagram with (n=8) is considered for the turbo encoder as revealed in Figure 2. Initially the state of the turbo encoder is (S0=‘0’) and S(n) ϵ {0, 1… 7} are the subsequent states for S(n) ϵ {0, 1} message bit. The states are S0(000), S1(001), S2(010), S3(011), S4(100), S5(101), S6(110), and S7(111) corresponding to binary message inputs (di=0, di=1). The interleaver is the main block in the turbo encoding and decoding process. Due to the interleaving process burst errors in consecutive bits are spread over the transmitted sequence. Thus, bit error rate performance is improved as the bits are reordered in some pseudo-random manner. Another process is puncturing, which increases the code rate by removing the coded parity bits of the sequence thus, making the channel utilization efficient. The interleaver in the turbo encoder structure generates a long data block by reshuffling the data and the interleaver in the decoder structure corrects the errors by bypassing the data to the first decoder. After this process, the interleaver further reshuffles the firstly decoded data and once
- 3. ISSN: 2089-4864 Int J Reconfigurable & Embedded Syst, Vol. 12, No. 2, July 2023: 174-185 176 shuffling is done passes the data to the second decoder for remaining error corrections. The same way process is repeated. The organization of the manuscript is as follows. Section 2 offerings the related work with the identification of the gaps, problem statement, and originality. Section 3 offerings the proposed method. Section 4 details the methodology and implementation strategy followed by parallel decoding and interleaving design architecture. Section 5 offerings the results and deliberations part with the possible simulation test inputs and analysis for the FPGA parameters followed by conclusions in section 6. Figure 1. Turbo encoder [11] Figure 2. Trellis diagram configuration [11] 2. RELATED WORK Turbo codes are used in 4G mobile infrastructures as well as in deep-space satellite communications and other applications where inventors want to achieve dependable data transmission over bandwidth-or latency-inhibited transmission links in the existence of data-corrupting noise. The hardware application of a serial turbo decoder architecture [12] was done on FPGA for LTE and worldwide interoperability for microwave access (Wi-MAX) communication system. An interleaver memory (ILM) was used for LTE systems that utilized comparators and subtractors instead of multipliers and dividers. This unique interleaver overcomes the need for memory utilization as compared to the parallel architecture approach. The algorithm was proposed for the scheme of the interleaver [13] for turbo decoder architecture. The error correction performance, as well as latency, was improved in comparison to the sequential decoder by the integration of the algorithm. The algorithm involves the parallel architecture structure which is highly efficient in relations of area, and BER thus, improving the throughput respectively. Very high-speed integrated circuit hardware description language (VHDL) was used for the application of different error control coding [14] encoders and decoders on FPGA. The error control codes is
- 4. Int J Reconfigurable & Embedded Syst ISSN: 2089-4864 Turbo encoder and decoder chip design and FPGA device analysis for … (Aakanksha Devrari) 177 an important area of discussion in telecommunication scenario since it provides efficient ways of error correction and error detection. The analysis of different channel coding schemes [15] was proposed for the fifth-generation (5G) mobile system based on binary phase-shift keying (BPSK) modulation over Additive white Gaussian noise (AWGN) channels such as polar, turbo, and low-density parity-check (LDPC). In terms of computational complexity, polar codes proved to be the best of all. A turbo decoder architecture was presented based on the log MAP algorithm [16] using a parallel processor to meet the requirements of long- term evolution communication standards. It is observed that this structure gives better performance in relation to BER. The synthesis results show the device utilization such as the count values of slices and 4 input lookup tables (LUTs) is less using this algorithm, making the design cost-effective. Based on the study and discussion on the forward error correction schemes [17] such as turbo codes and conventional coding, it is observed that turbo codes depict the best theoretical bound in terms of efficiency and channel capacity. In recent times these codes are preferred in several applications such as CDMA-2000 and 3G cellular networks such as UMTS. The log-MAP algorithm was used for turbo decoding [18] in the Wi-Max communication system. The lookup table (LUT) based log-MAP algorithm was proposed for turbo decoder structure in LTE wireless communication system [19] that reduces the logic resource utilization by 15% in comparison to the conventional design approach. The implementation of a concurrent turbo decoder [20] with an inverse address originator scheme was proposed for future broadcasting communication systems. This address- generation scheme reduces the iteration processing time which further reduces latency. Xilinx integrated logic analyzer (ILA) tool is used to check the validity of the turbo decoder, block error rate (BLER), and latency. A new design approach for the realization of a turbo decoder on [21] FPGA with a reverse address generator to reduce the complexity and latency. To reduce memory issues a contention-free interleaver is designed and a clock gating technique is used to improve power dissipation. The turbo decoder base was proposed to get high throughput. This was achieved using a processor in which the algorithm has accelerator units and highly controlled software. As the need for configurable memory interfacing is required memory interfacing with a switch fuse unit is used instead of fixed memory interfacing. The parallel decoder [22] based on configurable interleaved networking was used for a long-term evolution standard. A multi-parallelism interleaver approach [23] was designed to reduce computational complexity. The performance loss due to parallel structure is compensated using path metrics [24]. The proposed design is a competitive solution to recent work in terms of high reliability and efficiency. The related work presents the turbo decoder design and FPGA implementation with different aspects. They have implemented the LTE turbo decoder with the Max log-MAP technique on Virtex-6 XC6VLX75T with a clock frequency of 270 MHz, the number of slices 4108, and LUTs 6310. The work is also done on the VHDL-based design to implement a turbo decoder with a log-MAP technique that follows the decoding iterations but not analyzed FPGA resource utilization. The performance of the turbo codec is also estimated with the log-MAP algorithm on central processing unit (CPU) and FPGA and further utilization of the FPGA for LTE. The work is not done on the latest FPGA that provides fast switching, minimum FPGA hardware parameters utilization, memory, and delay. The problem declaration of the study is to analyze the performance of the hardware chip design of the turbo encoder and decoder based on simple architecture by the integration of the log-MAP algorithm in the same chip, perform the Xilinx simulation for the data verification, and compare the chip design performance on latest supporting FPGA with hardware and timing parameters. The originality of the work is that the encoder and decoder design supports greater frequency with optimal timing and hardware FPGA resources. 3. PROPOSED ALGORITHM The decoder structure is a parallel concatenated arrangement of convolution codes that takes the independent set of parity bits. After a systematic bit sequence is formed this process is known as the iterative decoding method [25]. The iterative decoding mechanism and the configuration of the turbo decoder are shown with the help of Figure 3 and Figure 4. The information or the message bits are processed between soft input and soft-output (SISO)-1 to SISO-2. Extrinsic information generated by SISO-1 is de-interleaved and fed as input to the next SISO-2. This building block calculates the forward matrix value (α), backward matrix value (β), and the output SISO matrix [5]. To calculate α and β values the Bahl, Cocke, Jelinek, and Raviv (BCJR) algorithm is applied. The SISO decoders perform the turbo decoding process and are referred to as a log-likelihood ratio (LLR) [26]. For the encoded sequence, di = [d1, d2…dN], and generated codeword against each sequence di = [d1, d2…dN], the LLR is articulated as (1).
- 5. ISSN: 2089-4864 Int J Reconfigurable & Embedded Syst, Vol. 12, No. 2, July 2023: 174-185 178 Figure 3. Turbo decoding [5] Figure 4. Iterative turbo decoding [26] 𝐿(𝑑𝑖) = ln 𝑃(𝑑𝑖=1) 𝑃(𝑑𝑖=0) (1) For the decoded sequence h=[h1, h2…hN], 𝐿(𝑑𝑖|ℎ) = ln 𝑃(𝑑𝑖=1|ℎ) 𝑃(𝑑𝑖=0|ℎ) (2) Bahl et al. [1] discussed the MAP algorithm in 1974, which was applied to breed the probabilities of individual bits and descend the extrinsic information. The LLR of the kth symbol is given as (3). 𝐿(𝑑𝑘|ℎ) = 𝑙𝑛 ∑ 𝛼𝑘−1 ′ 𝑑𝑘=1 (𝑠𝑘−1)𝛾𝑘 ′(𝑠𝑘−1,𝑠𝑘)𝛽𝑘 ′ (𝑠𝑘) ∑ 𝛼𝑘−1 ′ 𝑑𝑘=0 (𝑠𝑘−1)𝛾𝑘 ′(𝑠𝑘−1,𝑠𝑘)𝛽𝑘 ′ (𝑠𝑘) (3) In the (3): 𝛼𝑘 ′ =Forward state matric 𝛽𝑘 ′ =Backward state matric 𝛾𝑘 ′ =Branch matric 𝑠𝑘=Trellis diagram states at the time ‘k’ The MAP algorithm [26] comprises addition processes and multiplication processes. For a large classification, the logarithm and estimate are applied, and the (3) for Max log-MAP is specified as (4). ln(∑ 𝑒𝑔𝑖) ≈ 𝑚𝑎𝑥𝑖𝑚𝑢𝑚 (𝑔𝑖) (4) Then the algorithm is streamlined to the (5). 𝐿(𝑑𝑘|ℎ) = 𝑚𝑎𝑥𝑑𝑘=1(𝛼𝑘−1 𝛾𝑘(𝑠𝑘−1, 𝑠𝑘) 𝛽𝑘 (𝑠𝑘)) − 𝑚𝑎𝑥𝑑𝑘=0(𝛼𝑘−1 𝛾𝑘(𝑠𝑘−1, 𝑠𝑘) 𝛽𝑘 (𝑠𝑘)) (5) The expression has addition operations and association operations. The state matrix is similarly condensed and written as (6) and (7). 𝛼𝑘 (𝑠𝑘) = ln( 𝛼𝑘 ′ (𝑠𝑘)) = 𝑚𝑎𝑥𝑖𝑚𝑢𝑚 (𝛼𝑘−1(𝑠𝑘−1) + 𝛾𝑘−1(𝑠𝑘−1)) (6)
- 6. Int J Reconfigurable & Embedded Syst ISSN: 2089-4864 Turbo encoder and decoder chip design and FPGA device analysis for … (Aakanksha Devrari) 179 𝛽𝑘 (𝑠𝑘) = ln( 𝛽𝑘 ′ (𝑠𝑘)) = 𝑚𝑎𝑥𝑖𝑚𝑢𝑚 (𝛽𝑘−1(𝑠𝑘−1) + 𝛾𝑘−1(𝑠𝑘−1)) (7) The branch matrix is expressed as (8). 𝛾𝑘(𝑠𝑘−1, 𝑠𝑘) = { 𝛾𝑘00 = 0.5 (𝐿𝑎(𝑑𝑘) + 𝐿(𝑑𝑖𝑛) + 𝐿(𝑃𝑘1) 𝛾𝑘01 = 0.5 (𝐿𝑎(𝑑𝑘) + 𝐿(𝑑𝑖𝑛) − 𝐿(𝑃𝑘1) 𝛾𝑘10 = 0.5 (−𝐿𝑎(𝑑𝑘) − 𝐿(𝑑𝑖𝑛) + 𝐿(𝑃𝑘1) 𝛾𝑘11 = 0.5 (−𝐿𝑎(𝑑𝑘) − 𝐿(𝑑𝑖𝑛) − 𝐿(𝑃𝑘1)} (8) Here, 𝐿𝑎=Priori LLR 𝑑𝑖𝑛=Methodical bit to the realization of extrinsic information 𝑃𝑘1=Bit for parity check Subsequently, after processing the calculations of the LLR, the resultant extrinsic information is presented as (9). 𝐿𝑒(𝑑𝑘) = 𝜇. (𝐿(𝑑𝑘|ℎ) − 𝐿(𝑑𝑖𝑛) − 𝐿𝑎(𝑑𝑘)) (9) Where μ denotes the scaling factor, and for accessing the extrinsic information in the MAP procedure to recompense for the losses, μ<1. 4. METHOD To meet the decoding requirements in the turbo coding scheme, interleaver and de-interleaver blocks are placed in parallel. The most challenging factor in the interleaving process is the contention of memory. Figure 5 shows the interleaving and parallel processing of the turbo decoder. To extract extrinsic information in terms of forwarding and backward variables, successive interleaving is used. Depending on the block size of data-parallel execution is used. An address generator unit (AGU) and a control unit (CU) are used to map the interleaver and de-interleaved with random access memory (RAM). Figure 5. The execution and parallel realization of the interleaving concept in the turbo hardware decoder The real-time log-likelihood ratio and the parallel extrinsic information are selected with the help of multiplexers to perform the interleaving process. To perform the sequential read and write operations control signals are linked with RAM. To improve latency and reduce processing time during each iteration the AGU
- 7. ISSN: 2089-4864 Int J Reconfigurable & Embedded Syst, Vol. 12, No. 2, July 2023: 174-185 180 is used. For the selection of SISO and first input first output (FIFO) modules switch matrices are used. Depending upon the priority, each FIFO module is attached to its current SISO module for input-output buffer synchronization. Since, the LLR, branch matrices, forward matrices, and backward matrices parameters are sequential. The throughput of the turbo decoder can be maximized using input and output buffers [27]. The design methodology of the work is as follows. The chip functionality follows the steps of the chip design, and simulation modeling. The VHDL programming is used to design the turbo encoder and decoder chip that accepts the 32-bit input stream. The waveform simulation is carried out for both the integrated decoder module in which the test cases are verified. The real-time verification is done on the Virtex-5 FPGA board to check the feasibility of the design. The method for the same is given in Figure 6. Figure 6. Method 5. RESULTS AND DISCUSSIONS In Xilinx ISE 14.7, the behavioral modeling-based model is used to develop the turbo encoder and decoder. The encoder and decoder integrated chip register transfer level (RTL) view are shown in Figure 7. Table 1 lists the specifics of RTL pins. Figure 8 depicts the waveform in Xilinx ISIM for the decoder for test- 1 and test-2 in binary without error in reception. Figure 9 depicts the waveform in Xilinx ISIM for the decoder for test-1 and test-2 in hexadecimal without error in reception. Figure 10 depicts the waveform in Xilinx ISIM for the decoder for test-3 and test-4 in binary with an error in reception. Figure 11 depicts the waveform in Xilinx ISIM for the decoder for test-3 and test-4 in hexadecimal with an error in reception. The major inputs and outputs of the decoder's main stimuli are displayed on the simulation test bench. Table 1. Pin description of encoder module and decoder module RTL chip Pins Direction Description Clk Input The clock signal has a 50% duty cycle and is an input clock signal that can provide a rising edge-based clock signal. Reset Input To maintain the contents 0 for the encoder and decoder, reset is the chip design's default input. Turbo_tx_data_in_encoder [31:0] Input It gives the turbo encoder provided at the transmitter section of the design the 32-bit input message. Turbo_rx_data_out_decoder [31:0] Output The 32-bit turbo-encoded output can be treated as an AWGN channel's input. Turbo_enc_data_inout [31:0] Inout It is the 32-bit output of the AWGN communication channel and feedback to the turbo decoder. Error Output It is the indicator output that a correct or incorrect message is received in the decoder section.
- 8. Int J Reconfigurable & Embedded Syst ISSN: 2089-4864 Turbo encoder and decoder chip design and FPGA device analysis for … (Aakanksha Devrari) 181 Figure 7. RTL view of the integrated turbo encoder and decoder Figure 8. Waveform in Xilinx ISIM for decoder for test-1 and test-2 in binary Figure 9. Waveform in Xilinx ISIM for decoder for test-1 and test-2 in hexadecimal Figure 10. Waveform in Xilinx ISIM for decoder for test-3 and test-4 in binary Figure 11. Waveform in Xilinx ISIM for decoder for test-3 and test-4 in hexadecimal
- 9. ISSN: 2089-4864 Int J Reconfigurable & Embedded Syst, Vol. 12, No. 2, July 2023: 174-185 182 Test case (turbo encoder/ decoder): − Apply Clk=rising edge clock signal, reset=‘0’, turbo_tx_data_in_encoder [31:0]=“0000 0000 0000 0000 1100 1100 0000 0000” (binary)=0000CC00 (hexadecimal), then turbo_enc_data_inout [31:0]=“0000 0000 0000 0000 1100 1100 0000 0000” (binary)=0000CC00 (hexadecimal), and turbo_rx_data_out_decoder [31:0]=“0000 0000 0000 0000 1100 1100 0000 0000” (binary)=0000CC00 (hexadecimal), with error=‘0’ − Apply Clk=rising edge clock signal, reset=‘0’, turbo_tx_data_in_encoder [31:0]=“0000 0000 0000 0000 1010 1111 0000 0000” (binary)=0000AF00 (hexadecimal), then turbo_enc_data_inout [31:0]=“0000 0000 0000 0000 1010 1111 0000 0000” (binary)=0000AF00 (hexadecimal), and turbo_rx_data_out_decoder [31:0]=“0000 0000 0000 0000 1010 1111 0000 0000” (binary)=0000AF00 (hexadecimal) with error = ‘0’ − Apply Clk=rising edge clock signal, reset=‘0’, turbo_tx_data_in_encoder [31:0]=“0000 0000 0000 0000 1111 0010 0000 0000” (binary)=0000F200 (hexadecimal), then turbo_enc_data_inout [31:0]=“0000 0000 0000 0000 1111 0001 0000 0000” (binary)=0000F100 (hexadecimal), and turbo_rx_data_out_decoder [31:0]=“0000 0000 0000 0000 1111 0010 0000 0000” (binary)=0000F200 (hexadecimal) with error=‘1’ − Apply Clk=rising edge clock signal, reset=‘0’, turbo_tx_data_in_encoder [31:0]=“0000 0000 0000 0000 0011 1111 0000 0000” (binary)=00003F00 (hexadecimal), then turbo_enc_data_inout [31:0]=“0000 0000 0000 0000 0010 1111 0000 0000” (binary)=00002F00 (hexadecimal), and turbo_rx_data_out_decoder [31:0]=“0000 0000 0000 0000 0011 1111 0000 0000” (binary)=00003F00 (hexadecimal) with error=‘1’. Table 2 summarizes the hardware parameters of the turbo encoder and decoder hardware with MAP algorithm in Xilinx ISE 14.7 software for various parameters targeting the Virtex-5 FPGA. Slices, flip-flops, LUTs, and input/output block (IOBs) are the hardware parameters [28]. Table 3 shows the results of timing simulations for the turbo encoder and decoder chip hardware for time-concerned parameters such as minimum period (ns), minimum and maximum duration before and after the clock signal (ns), and maximum supporting frequency [29]. Table 2. Summary of FPGA hardware usage Hardware Turbo codes Turbo codes with Max log-MAP Encoder Decoder Encoder Decoder Slices 75 150 67 140 Flip-flops 87 103 77 80 LUTs 120 125 103 107 IOBs 64 96 64 96 Global Clocks (GCLKs) 1 1 1 1 Table 3. Parameters connected to timing information Timing parameter Turbo codes Turbo codes with Max log-MAP Encoder Decoder Encoder Decoder Frequency (MHz) 314.0 320.0 332.00 355.0 Minimum period (ns) 1.710 1.420 1.6120 1.390 Time (minimum) before clock (ns) 2.109 2.917 1.964 2.851 Time (maximum) after clock (ns) 4.320 5.325 4.124 4.109 Combinational path delay (ns) 8.139 9.662 7.700 8.350 Speed grade -5 -5 -5 -5 Figure 12 presents the FPGA hardware utilization graph for the turbo encoder and decoder with a simple and integrated Max log-MAP algorithm chip. Figure 13 presents the time-related parameters of FPGA hardware. The combinational path delay and related time parameters are reduced by the integration of the Max log-MAP algorithm in the turbo encoder and decoder hardware due to parallel execution. The major motivation for parallel implementation is to efficiently execute code, as it saves time and allows programmers to be executed in a shorter clock time. There is significantly less disruption for workers and equipment because multiple processes can run simultaneously. The encoder and decoder both utilize less hardware resources in comparison to the turbo encoder and decoder due to the parallel execution of the block. The designed hardware chip reduces the hardware utilization in the FPGA synthesis.
- 10. Int J Reconfigurable & Embedded Syst ISSN: 2089-4864 Turbo encoder and decoder chip design and FPGA device analysis for … (Aakanksha Devrari) 183 Figure 12. FPGA hardware usage Figure 13. FPGA timing parameters 6. CONCLUSION The hardware chip design of the turbo encoder and decoder is done successfully in Xilinx ISE 14.7. The Max log-MAP algorithm is integrated with the chip. The comparative performance of the chip is studied based on hardware and timing parameters on Virtex-5 FPGA synthesis. The ISIM simulation waveform shows the successful decoding of the data using the turbo decoder. The turbo encoder with Max log-MAP utilizes less number of slices (67), flip-flops (71), and LUTs (103) in comparison to the turbo encoder. In the same way, the turbo decoder with Max log-MAP utilizes less number of slices (140), flip-flops (80), and LUTs (107) in comparison to the turbo decoder. The frequency of the turbo encoder and decoder is 314 MHz and 320 MHz respectively. The frequency of the turbo encoder and decoder with the Max log-MAP algorithm is 332 MHz and 350 MHz respectively. The higher frequency support indicates that the turbo encoder and decoder with the Max log-MAP algorithm provide the faster response in comparison to the turbo encoder and decoder. The combination path delay of the turbo encoder and decoder with the Max log-MAP algorithm is 7.700 ns and 8.350 ns respectively, which is less in comparison to the combinational path delay of the turbo encoder and decoder. The designed encoder and decoder hardware chip with the Max log-MAP algorithm provides less hardware utilization, higher frequency support, and less delay in comparison to the turbo encoder and decoder.
- 11. ISSN: 2089-4864 Int J Reconfigurable & Embedded Syst, Vol. 12, No. 2, July 2023: 174-185 184 REFERENCES [1] L. Bahl, J. Cocke, F. Jelinek, and J. Raviv, “Optimal decoding of linear codes for minimizing symbol error rate,” IEEE Transactions on Information Theory, vol. 20, no. 2, pp. 284–287, Mar. 1974, doi: 10.1109/TIT.1974.1055186. [2] C. Berrou and A. Glavieux, “Near optimum error correcting coding and decoding: turbo-codes,” IEEE Transactions on Communications, vol. 44, no. 10, pp. 1261–1271, 1996, doi: 10.1109/26.539767. [3] R. Mishra, P. Kuchhal, and A. Kumar, “Effect of height of the substrate and width of the patch on the performance characteristics of microstrip antenna,” International Journal of Electrical and Computer Engineering (IJECE), vol. 5, no. 6, pp. 1441–1445, Dec. 2015, doi: 10.11591/ijece.v5i6.pp1441-1445. [4] K. H. Kaittan and S. J. Mohammed, “Implementing and designing a secure information system based on the dsss gold sequence using matlab,” Bulletin of Electrical Engineering and Informatics, vol. 10, no. 3, pp. 1455–1463, Jun. 2021, doi: 10.11591/eei.v10i3.2446. [5] M. J. Adamu, L. Qiang, R. S. Zakariyya, C. O. Nyatega, H. B. Kawuwa, and A. Younis, “An efficient turbo decoding and frequency domain turbo equalization for LTE based narrowband internet of things (NB-IoT) systems,” Sensors, vol. 21, no. 16, Aug. 2021, doi: 10.3390/s21165351. [6] C. E. Shannon, “A mathematical theory of communication,” Bell System Technical Journal, vol. 27, no. 3, pp. 379–423, 1948, doi: 10.1002/j.1538-7305.1948.tb01338.x. [7] L. Hang, G. Zhen, Z. Ming, and W. Jing, “Partial iterative decode of turbo codes for on-board processing satellite platform,” China Communications, vol. 12, no. 11, pp. 1–8, Nov. 2015, doi: 10.1109/CC.2015.7366233. [8] L. Fanari, E. Iradier, I. Bilbao, R. Cabrera, J. Montalban, and P. Angueira, “Comparison between different channel coding techniques for IEEE 802.11be within factory automation scenarios,” Sensors, vol. 21, no. 21, Oct. 2021, doi: 10.3390/s21217209. [9] P. Salija, B. Yamuna, T. R. Padmanabhan, and D. Mishra, “Generic direct approach for decoding turbo codes using probability density based reliability model,” Journal of Communications Technology and Electronics, vol. 66, no. 2, pp. 175–183, Feb. 2021, doi: 10.1134/S1064226921020133. [10] A. Devrari, A. Kumar, A. Kumar, and S. Singh, “Design and FPGA implementation of DSSS for near-far effect in MANET,” in Proceeding of International Conference on Intelligent Communication, Control and Devices . Advances in Intelligent Systems and Computing, Singapore: Springer, 2017, pp. 425–434. [11] L. F. Gonzalez-Perez, L. C. Yllescas-Calderon, and R. Parra-Michel, “Parallel and configurable turbo decoder implementation for 3GPP-LTE,” in 2013 International Conference on Reconfigurable Computing and FPGAs (ReConFig), Dec. 2013, pp. 1–6, doi: 10.1109/ReConFig.2013.6732316. [12] J. P. Woodard and L. Hanzo, “Comparative study of turbo decoding techniques: an overview,” IEEE Transactions on Vehicular Technology, vol. 49, no. 6, pp. 2208–2233, 2000, doi: 10.1109/25.901892. [13] R. Dobkin, M. Peleg, and R. Ginosar, “Parallel VLSI architecture for MAP turbo decoder,” in The 13th IEEE International Symposium on Personal, Indoor and Mobile Radio Communications, 2002, pp. 384–388, doi: 10.1109/PIMRC.2002.1046727. [14] J. M. Gilbert, C. Robbins, and W. Sheikh, “FPGA implementation of error control codes in VHDL: an undergraduate research project,” Computer Applications in Engineering Education, vol. 27, no. 5, pp. 1073–1086, Sep. 2019, doi: 10.1002/cae.22137. [15] Z. R. M. Hajiyat, A. Sali, M. Mokhtar, and F. Hashim, “Channel coding scheme for 5G mobile communication system for short- length message transmission,” Wireless Personal Communications, vol. 106, no. 2, pp. 377–400, May 2019, doi: 10.1007/s11277- 019-06167-7. [16] S. Kalyani, V. Raj, and K. Giridhar, “Narrowband interference mitigation in turbo-coded OFDM systems,” in 2007 IEEE International Conference on Communications, Jun. 2007, pp. 1059–1064, doi: 10.1109/ICC.2007.180. [17] J. D. Kene and K. D. Kulat, “Soft output decoding algorithm for turbo codes implementation in mobile Wi-Max environment,” Procedia Technology, vol. 6, pp. 666–673, 2012, doi: 10.1016/j.protcy.2012.10.080. [18] J. Li, X. Wang, J. He, C. Su, and L. Shan, “Turbo decoder design based on an LUT-normalized Log-MAP algorithm,” Entropy, vol. 21, no. 8, Aug. 2019, doi: 10.3390/e21080814. [19] H. Luo et al., “Low latency parallel turbo decoding implementation for future terrestrial broadcasting systems,” IEEE Transactions on Broadcasting, vol. 64, no. 1, pp. 96–104, Mar. 2018, doi: 10.1109/TBC.2017.2704425. [20] V. Akshaya, K. N. Sreehari, and A. Chalil, “VLSI implementation of turbo coder for LTE using verilog HDL,” in 2020 Fourth International Conference on Computing Methodologies and Communication (ICCMC), Mar. 2020, pp. 275–279, doi: 10.1109/ICCMC48092.2020.ICCMC-00051. [21] P. Salmela, H. Sorokin, and J. Takala, “A programmable max-log-MAP turbo decoder implementation,” VLSI Design, vol. 2008, pp. 1–17, Dec. 2008, doi: 10.1155/2008/319095. [22] Z. Yan, G. He, W. He, S. Wang, and Z. Mao, “High performance parallel turbo decoder with configurable interleaving network for LTE application,” Integration, vol. 52, pp. 77–90, Jan. 2016, doi: 10.1016/j.vlsi.2015.05.003. [23] G. Masera, G. Piccinini, M. R. Roch, and M. Zamboni, “VLSI architectures for turbo codes,” IEEE Transactions on Very Large Scale Integration (VLSI) Systems, vol. 7, no. 3, pp. 369–379, Sep. 1999, doi: 10.1109/92.784098. [24] C. Berrou, “The invention of Turbo codes: the ten-year-old Turbo codes are entering into service,” IEEE Communications Magazine, vol. 61, no. 2, pp. 6–13, Feb. 2023, doi: 10.1109/MCOM.2023.10048673. [25] J. Nadal and A. Baghdadi, “Parallel and flexible 5G LDPC decoder architecture targeting FPGA,” IEEE Transactions on Very Large Scale Integration (VLSI) Systems, vol. 29, no. 6, pp. 1141–1151, Jun. 2021, doi: 10.1109/TVLSI.2021.3072866. [26] A. A. Hamad, M. T. Gatte, and L. A. Abdul-Rahaim, “Efficient systematic turbo polar decoding based on optimized scaling factor and early termination mechanism,” International Journal of Electrical and Computer Engineering (IJECE), vol. 13, no. 1, pp. 629–637, Feb. 2023, doi: 10.11591/ijece.v13i1.pp629-637. [27] A. Kumar, P. Sharma, M. K. Gupta, and R. Kumar, “Machine learning-based resource utilization and pre-estimation for network on chip (NoC) communication,” Wireless Personal Communications, vol. 102, no. 3, pp. 2211–2231, Oct. 2018, doi: 10.1007/s11277-018-5376-3. [28] A. Kumar, G. Verma, and M. K. Gupta, “FM receiver design using programmable PLL,” Wireless Personal Communications, vol. 97, no. 1, pp. 773–787, Nov. 2017, doi: 10.1007/s11277-017-4536-1. [29] A. Kumar, A. Kumar, and A. Devrari, “Hardware chip performance analysis of different FFT architecture,” International Journal of Electronics, vol. 108, no. 7, pp. 1124–1140, Jul. 2021, doi: 10.1080/00207217.2020.1819441.
- 12. Int J Reconfigurable & Embedded Syst ISSN: 2089-4864 Turbo encoder and decoder chip design and FPGA device analysis for … (Aakanksha Devrari) 185 BIOGRAPHIES OF AUTHORS Ms. Aakanksha Devrari is B.Tech. in Electronics and Communication Engineering from Uttaranchal institute of Technology, (UTU), Dehradun India in 2011. M.Tech. in VLSI Design, Faculty of Technology, Uttarakhand Technical University (UTU), Dehradun India in 2013. Pursuing Ph.D. (Electronics Engineering) in the domain of VLSI design. Her areas of interest are VLSI design, digital communication system. She has published more than 10 research papers in international journals and conferences. She can be contacted at email: [email protected]. Dr. Adesh Kumar is working as Senior Associate Professor in the Department of Electrical & Electronics, Engineering ‘‘University of Petroleum and Energy Studies’, Dehradun, India. He is B.Tech in Electronics and Communication Engineering from Uttar Pradesh Technical University Lucknow, India in 2006. M.Tech. (Hons) in Embedded Systems Technology, from SRM University, Chennai in 2008. Ph.D. (Electronics Engineering) form University of Petroleum and Energy Studies (UPES), Dehradun India in 2014. He has also worked as Senior Engineer in TATA ELXSI LIMITED Bangalore and faculty member in ICFAI University, Dehradun. His areas of interest are VLSI design, embedded systems design, telecommunications and signal processing. He has published more than 100+ research papers in international peer reviewed journals (SCI/Scopus) and conferences. He is the reviewer of many SCI/SCIE and Scopus journals such as IEEE Transactions of Industrial Electronics, Wireless Personal Communications, Microsystem, Technologies, 3D research, Springer, Journal of Electronic Science and Technology, Journal of Visual Languages and Computing, Elsevier, World Journal of Engineering and Emerald are few of them. He has supervised 7 Ph.D scholars and 5 candidates are doing research under his supervision. He has chaired more than 10 sessions in international conferences and involved in the Ph.D. evaluation committee in UPES, India. Guest Editor: Special Issue "Intelligent Devices & Computing Applications" in Computer Systems Science & Engineering, Journal, Tech Science, 2022 (SCI Indexed). Guest Editor: Special Issue "Intelligent Communication and Smart Grid Automation Applications" in Intelligent Automation and Soft Computing, Journal, Tech Science, 2021 (SCI Indexed). Editor: Proceedings of Intelligent Communication, Control and Devices in Advances in Intelligent system and Computing (AISC), Book Series, Springer, Vol. 989, 2020. (Scopus Indexed). He can be contacted at email: [email protected] and [email protected].