ee 4
ar
TT
CONVOLUTIONAL Congr 5
—
output. The constraint lengths of
whose length is the numb
diagram,
the encoder form avector
Of inputs in the encoder
Q4 Design an encoder for the (7,4) binary cyclic
generated by g (x) I++ x' and verify is operas
using the message vectors (1001) and (1011).
MREU.Dee20i3p
es
‘Ans. Given g(x) = 1+ x+x°
‘The given generator polynomial can be written
as.
refine Code Tree. [R.T.U. 2018, 2016]
s)= 14 xt Oe
&
(Comparing with equation x + g.x°+g.x*1)
Encoder for generator polynomial g(x)= 1+ 3+
Define Trellis. IRTU, 2016] isgiven by figure
a atest
Ans. Trellis : —
(2) It indicates transitions from current to next state. at: .
(0) Itis shorter or compact way of representing coding [oi : am rf
process.
) Decoding is little complex using trellis diagram, cute
(@ Itrepeats in every stage in steady state, it has only eS
one stage. ‘codeword to
Ps ‘be transmitted
(¢) Itis simpler to implement in programming. ni
—O
5 ved vector Z
& Define Constraint Length. IRTU.20161 ) Qs For a (7, 4) cyclic code, the ee Ms Draw the
aera
SS ———_—_—_—___ ll
Ass. Constraint Length : Constraint length of a
‘ssvolutional code is defined as the number of shifts over
ich a single information bit can influence the encoder
=I
(X) is LITO101 and gO) 1 es the single
syndrome calculation ct and ibe 2013) @
‘Ans. Given generator polynomial is
BA=Ttx tx?
gx)ax tox tix]
‘We know general form of gene’
se 1
‘Comparing (i) & (ii) we set
: calculate :
Te
:
MT aes se
is
otynomial
tor po
(i)
yarexts!
coding and decoding in the convolution
re (RT. 2017]
OR
Deceing Probability of Convolution code
F Z is [RU 2018]
‘Ans. in telecommunication, a convolutional code is a type
oferor-comecting code that generates party symbols via
‘the sliding application of a boolean polynomial function to
lee llis decoding using
i trellis. Time invariant trellis decoding
rational eodesto be maximum-likelihood soft
ihreasonable complex I
onvolut S are often characterized by th
The ea the depth (or memory) ofligetene
ut daa rate a Ply Biven as nk, where
eS, eat Out symbol rate
callethe "constraint length yer a:
cuca tength"'K, where
We" described ag
Mvolutional
Jock codes typically employ termination, SN
block eodes SR
saath of convolutional codes cam algo}
Oe oa codes, which generally havea
uths that are determined by algebraic c
lenee code rate of a convolutional code sad
bol puncturing. For
modified via symbol pi
convolutional code with a mother’ code rate
perpunetured toa higher rate of, for example ga
nae jortion of code
sy not transmitting a portion i
fe fo mance of a pi nctured convolutional emote
performa
ount of parity trangm
scales well with the amount
ability to perform economical soft ansimited +1
on
lutional codes, as well as the block
parity check digits,
Interlaced Codes for Burst Error Correction
In general, random-error correcting codes are not
efficient for burst-error correcting and burst-error
correcting codes are not efficient for random-error
correcting. Out of the several methods proposed! to
simultaneoualy random and burst errors the interlaced code
is the most effective,
For an (n,k) code, if we interlace , code words,
which is known as a (7,4) interlaced code, instead of
transmitting codewords one by one, we group i code:
words and interlace them,
Consider for example, the case of \=3 ant a
error correcting (15,8) code, Each code
‘We group code words to be transmitted
‘Suppose that first three code words (0 |
AGaeagioo VOW MDS )ANT FEL!
[Js Jos [Pe
[>a [2 [os |= | ma
lala [als
; HeXedis 9 WrVa0~
AVM aFieAayPD ED, Dotted line => 9/P Mag
Where
Solid tine => YP My
er peo error correcting code, (WO
Tbe corrected. Hence all the
° general, ifthe original
Correcting, the interdaced code can
tursts of length A oF less.
ae
Ams Trelis codes : Refer 10 2
‘Trelis diagram
ie sit mgiser M, M,, will indicate the state of
State @ Branch wort
then it is most likely thate™ =1, ‘meaning thatthe First
sion error and henee
should correct he (BIR BRT aa
1M received sequence is exci a
sequence corresponding to somy ary
through coder, then we shay
corresponding input sequent’
corresponding to some particular so
sder shoven In fit. 2
[fle1#]
|e
te
Hey
|
BIT ca vonces are oquilly lkely «
If all input data se
decoder that chooses ¢ if
Fig.
If we find no exact correspond
Sssume that input sequence to be one.
Benerates the fewer bit discrepant
compared to received sequence.
x Now illustrate vite algorithm let uy
Here %=S,@8,
%=S,OS.@s,
Fig.2
P(c|@)=maxp(re;) fora,
here ris the received sequence and ¢, is one of the
Possible transmitted sequences, is called the maximum
Uielinood decoder. The conditional Probabilities P(ro,)
(functions. Note tha for the BSC
‘annel) the maximu
Me consider aj).
. Possib)
inp ting pin Paths ty
Sat In Ne® Beneratg Point e is (a) at K = 1, a0
i output
ince, actual output is 10,
crepancies = |
;put = 0 should again yield
input =0 ‘eid output
.¢ the corresponding set of received
0 number discrepancies = 0, The
screpancies shown in circles.
pp me
ae
Fig. 4
16. Here we notice for output sequence of path through
trellis corresponding to input bt stream consisting
of all 0” yields minimum no of discrepancies. With
such as result we would then decide that input
sequence was all zero’s, the received sequence as
is readily verified should be all's
17. Because of noise of received sequence = 10.00 10
(00 00, the coder would have correct 160 <0
18. As the number of paths are reduced =. memory
required is reduced.
3) 0
a ef
a
Serger
Fig. 3
fh state (¢) at K =3 we go from (a) at K=1
“13 Explain the operation of any convolutional
w (a) K =2 and then from (a) at K = 2 to (c) at 2 P
encoder with the help of block diagram.
X=3.In first path discrepaney = ay
iasecond path discrepancy = 2
a saa, Encoder for Convolutional Code 5 0 encoder
Suse that we have to move from state (8) 8 } fora convolution codes Shows Fig. In this case
to state (a) K = L. How let us assume that K=no. of shift registers
Fah fom state (b) at K = 4 to state (a) at K= Lis Be
7 = no, of modulo? e482
to. of bits in thegode Boek
mum discrepancy path
We have to choose mini
K=4.
mi state (a) at K = 1 to state (b) at = Coded
Input
data
1. = leith of np dat sre"
‘a
wel mT Lg! ‘0001 | 000 | 000 | 000 | 10%
“= as | o10| 000 | 000 | 101,110
. y IT 000 | 000 | 101 | OIF
oe x | 5100-[ 000-| 101 | 110-| ort
‘i O10 000 | 101 | 110 | 110
I oll 000 | 10
+ Code oviput b, ot ort
“1001, 101 [ 110
1010 101
[ro11_| 101 | 110
1100_| 101 | O11
tio | 101 | on
i110_| 101 | 011
1itt_| 101 | 011
Pie Brcoter for Convlatonal Code
he output v, vp and v, of the adders are
s @ &)
®
5 es)
Irisassumed that, initially, the shift register is clear.
‘The operation of the encoder is explained for the input Table gives the coded output bi
‘data stream of a 4-bit sequence y Amie
ae data streams for the encoders shown in Feta ms
a column of input data stream is such that it is divides
i na ese |MSB. Thus, ‘two subsets (eight 0’s and eight | ’s), resultingin tWosbe
LS,=1,S;=0, $,=0. ofthe first cod its in the cad
Now. ¥l, v2and v3 can be found from Eqs. Thus ‘Oot ag co tare
%=180=1,v,=000=0,
4=180=1
Output at the frst-bit interval i
1 is
second bitinterval, $, = |, a4
‘Manner, outp i
Ste Ld nd my ete
ouput ateah pr ect = 443 = 7) pig tester
¥=3), Thus, foreach a Sit of y ane th The
int Ptteode word. Nojoe ge eV! ‘oh
"ise fork 8 Message itr
m4 sf ig onvoluti
VIR: OH oy from the Msp y
tree structure «nt NOt from ;
“we extremely eal fot