improved low complexity.pptx
TRANSCRIPT
-
7/29/2019 Improved Low Complexity.pptx
1/21
Submitted By
ANKIT SANCHETI
2011PEC5324
-
7/29/2019 Improved Low Complexity.pptx
2/21
TURBO Convolutional Codes (TCC) are also known as Parallel
Concatenated Convolutional Codes.
TCC have shown near Shannon capacity performance.
At BER of 105, in a 16-state rate-1/2 TCC with interleaver
length of 65536, /is 0.5 dB away from Shannons limitin
AWGN channel.
TCC decoder algorithm is computationally complex
-
7/29/2019 Improved Low Complexity.pptx
3/21
Concatenated coding schemes, using relatively simple constituent
Convolutional and Block codes, can achieve performance close to the
theoretical limits.
Concatenated Zigzag codes are low complexity parallel concatenated
block codes.
These codes perform better for large interleaver length. However, an
interleaver with large length has more time latency as each decoder
requires longer delay in interleaving and deinterleaving the received bits.
-
7/29/2019 Improved Low Complexity.pptx
4/21
BER performance of ILCHTC is better than that of LCHTC as error
convergence of ILCHTC is better than that of LCHTC.
ILCHTC decoder requires less number of iterations than LCHTC
decoder.
Decoding complexity of ILCHTC is considerably lower than that of
TCC.
For the interleaver length of N, trellis length for TCC decoder is N
times the number of constituent codes. For ILCHTC, maximum trellis
length is N
-
7/29/2019 Improved Low Complexity.pptx
5/21
-
7/29/2019 Improved Low Complexity.pptx
6/21
A. Constituent Encoder
A sequence of information bits is arranged in a
rectangular array of size . Alternatively, an (+((1))) information bit is
denoted by (, ).
d = {(, )}, 1 1 The Zigzag parity vector z(m)of constituent
encoder is,
z(m)= { ()()}, 1
-
7/29/2019 Improved Low Complexity.pptx
7/21
Zigzag parity bits are computed progressively as follows:
()(0) = 0
()( ) =[ (, ) + ()( 1)]2,
where 1
First, L rows of the array of information bits are encoded
using a rate-1/2 RSC code. Parity vector of RSC for row
of information array in first constituent encoder is,
J
j 1
-
7/29/2019 Improved Low Complexity.pptx
8/21
Interleaver for ILCHTC is designed in 2-stages.
1. Random interleaver is used to interleave N information bits.
2. N information bits are arranged in array.
-condition is realized in each of the J rows of array.
-condition ensures that the short distance between two errors is
mapped to the long distance by keeping at least distance
between the columns of() and() of array. If < Then,() () > , 0 <
-
7/29/2019 Improved Low Complexity.pptx
9/21
-
7/29/2019 Improved Low Complexity.pptx
10/21
Parallel concatenation ofconstituent encoders forms the
overall ILCHTC encoder.
Let represent the interleaverfor th constituent encoder.
Overall ILCHTC encoder with =
The transmitted code word, c is represented by,
c = {d, r(1), z(1), z(2), . . ., z(M)}
Code rate, of ILCHTC is given by
=/( + +)
-
7/29/2019 Improved Low Complexity.pptx
11/21
Soft In Soft Out A Posteriori Probability (APP)is used for decoding Zigzag-Convolutional
codes of first constituent decoder. A priori values of Log Likelihood Ratio (LLR) of
received information bits are computed.
LLRs are arranged in an array of size .
-
7/29/2019 Improved Low Complexity.pptx
12/21
i. Decode each row of the array using a priori LLRs as input to
Convolutional decoder. Output produced is a posteriori LLR of each
row of the information bits.
ii. A damping factor, D, is applied to a posteriori LLR found in step
(i). This ensures that LLRs are successfully updated in a gradual
way.
iii. Decode each column of the array using, output of (ii) as a priori
LLRs for Zigzag decoder.
iv. For overall decoder, global iterative decoding is implemented.
-
7/29/2019 Improved Low Complexity.pptx
13/21
Computational complexity of ILCHTC decoder depends upon the number
of multiplications and additions required to decode information bits.
The total trellis length is forTCC.
Let
be the number of states in Convolutional encoder,
be the number of Multiplications/Information Bit/Iteration (M/IB/I)
be the number of Additions/Information Bit/Iteration (A/IB/I)
required by a decoder.
-
7/29/2019 Improved Low Complexity.pptx
14/21
For Turbo Convolutional decoder,
= 8
= [(16+ 2) ] 2 For ILCHTC decoder,
= [ (8 4)]/
= [(16 1)/] +(5 + 4/) 1
For LCHTC decoder,
= [ (8 4)]/
= [(16 1)/] +(5 + 4/) 1
-
7/29/2019 Improved Low Complexity.pptx
15/21
-
7/29/2019 Improved Low Complexity.pptx
16/21
ILCHTC are simulated to investigate the BER performance at
various/with N = 1452.
An 8-state RSC code with generator polynomial,= [1,
(15/13)]8 is used in Convolutional encoders of TCC, LCHTC
and ILCHTC.
It is observed that ILCHTC achieves BER of 105 at/of
1.9 dB, which is 0.4 dB more than that for TCC.
-
7/29/2019 Improved Low Complexity.pptx
17/21
-
7/29/2019 Improved Low Complexity.pptx
18/21
-
7/29/2019 Improved Low Complexity.pptx
19/21
TCC requires less number of iterations than ILCHTC
and LCHTC.
In one iteration, the number of computations required
by TCC decoder is almost two times the number of
computations for ILCHTC and LCHTC decoders.
Therefore, overall decoder complexity of ILCHTC is
less than that of TCC.
-
7/29/2019 Improved Low Complexity.pptx
20/21
Complexity analysis shows that ILCHTC decoder requires less number of
computations than TCC decoder.
Rate-1/3 ILCHTC code requires 50% multiplication and 60% addition
equivalent operations per information bit per iteration as compared to
TCC.
For the interleaver length ofN = 1452, BER of 105 is achieved at /=
1.9 dB for ILCHTC which is 0.4 dB more than that for TCC.
Since TCC requires 10% less number of iterations and 50% more
computations/per iteration than ILCHTC the computational complexity
of overall ILCHTC decoder is 45% less than that of TCC decoder.
-
7/29/2019 Improved Low Complexity.pptx
21/21
Improved Low Complexity Hybrid Turbo Codes and their
Performance Analysis, by Archana Bhise and Prakash D.
Vyavahare, Member, IEEE
IEEE TRANSACTIONS ON COMMUNICATIONS, VOL. 58, NO.
6, JUNE 2010