Explicit evaluation of Viterbi's union bounds for the first event
error probability and the bit error probability of a binary
convolutional code on a binary symmetric channel
Citation for published version (APA):Post, K. A. (1976). Explicit evaluation of Viterbi's union bounds for the first event error probability and the bit error probability of a binary convolutional code on a binary symmetric channel. (Eindhoven University of Technology : Dept of Mathematics : memorandum; Vol. 7607). Technische Hogeschool Eindhoven.
Document status and date: Published: 01/01/1976
Document Version:
Publisher’s PDF, also known as Version of Record (includes final page, issue and volume numbers)
Please check the document version of this publication:
• A submitted manuscript is the version of the article upon submission and before peer-review. There can be important differences between the submitted version and the official published version of record. People interested in the research are advised to contact the author for the final version of the publication, or visit the DOI to the publisher's website.
• The final author version and the galley proof are versions of the publication after peer review.
• The final published version features the final layout of the paper including the volume, issue and page numbers.
Link to publication
General rights
Copyright and moral rights for the publications made accessible in the public portal are retained by the authors and/or other copyright owners and it is a condition of accessing publications that users recognise and abide by the legal requirements associated with these rights. • Users may download and print one copy of any publication from the public portal for the purpose of private study or research. • You may not further distribute the material or use it for any profit-making activity or commercial gain
• You may freely distribute the URL identifying the publication in the public portal.
If the publication is distributed under the terms of Article 25fa of the Dutch Copyright Act, indicated by the “Taverne” license above, please follow below link for the End User Agreement:
www.tue.nl/taverne
Take down policy
If you believe that this document breaches copyright please contact us at:
openaccess@tue.nl
providing details and we will investigate your claim.
TECHNISCHE HOGESCHOOL EINDHOVEN
Onderafdeling der Wiskunde
Memorandum 1976-07 mei 1976
EXPLICIT EVALUATION OF VITERBI'S UNION BOUNDS FOR THE FIRST EVENT ERROR PROBABILITY AND THE BIT ERROR PROBABILITY OF A BINARY CONVOLUTIONAL CODE
ON A BINARY SYMMETRIC CHANNEL
Technische Hogeschool Onderafdeling der Wiskunde PO Box 513, Eindhoven Nederland
by
EXPLICIT EVALUATION OF VITERBI'S UNION BOUNDS FOR THE FIRST EVENT ERROR PROBABILITY AND THE BIT ERROR PROBABILITY OF A BINARY CONVOLUTIONAL CODE
ON A BINARY SYMMETRIC CHANNEL
by
K.A. Post
Abstract. An explicit method is given to evaluate Viterbi's union bounds [IJ on both the first event error probability and the bit error probability of binary convolutional codes on a BSC. These bounds are explicitly given f or t e rate h I ' 2 code w1th generators 1 + D + D and 2 2 . 1 + D . Compar1son is made with bounds and experimental results of Van de Meeberg [2J.
I. Let p and q be positive num~ers, p <
!,
p + q = 1. Following Viterbi [IJ we'define a sequenceby the formulas ( ) ) k-) \ 2k-) 2k-l-j j /. ( J')p q j=O (k = 1,2,3, ... ).
Using the well-known addition property 1n Pascal's triangle we find (p + q)P2k -1 = P2k, so that P2k = P2k-1 (k ::: 1,2,3, .•. ) and furthermore so that (p + q)[P _ I
e
2k ) k kJ 2k :2 k P q 2k k k ::: P 2k+ I - P ( k ) P q , P 2k+ 1 = P 2k _ (1 _ 2 P )(2k) k k k P q (k = 1, 2 • 3 , ... ) •- 2
-Combining these results we obtain
I
o -
0 2 P3 P4 = :1-
p)[<O) + ( 1 ) pq] Io -
0 2<i)p2q2]
Ps
P6 = :1-
p)[ (0) + (j)pq + P = P == 1 _o -
p)[(O) 0 + (1) 2 pq + (4) 2 2(~)p\3J
2 P q + 7 8 2...
"...
"..
It is a well-known fact that for complex
z, Izi
<!
(2) (I - 4z)
(0)
°
+(2)
I Z +(4) 2
2 Z +(6) 3
3 z +so that the generating function F for the sequence P
1,P2,P3"" must read as follows (3) 00 F(z) :=
I
Pkzk k=) z 2 _1 - - [~ - (~ - p) (1 - 4 pq z ) 2] • - Z _1The Taylor-series in this formula converges for complex z,
Izl
< (4pq) 2 since z=
I is a removable singularity of the function F.n(z) 2. Let G be a
rational
function of the complex variable z, G(z) = d(z) ,where nand d are polynomials having gcd one, and nCO)
=
0.G has a Taylor-expansion around z
=
0;(4)
G(z)For our purpose we are interested 1n finding an explicit form for the expression
3
-Recall that the polynomial d can be uniquely factorized
(5)
where C is a complex constant and a. are the distinct zeros of d(z) with
J mu 1 tip 1 i ci tie sm. (j
=
1, ••• , r) .J
Then G(z) can be decomposed uniquely into partial fractions
(6) G(z)
m. r J
q(z) +
I I
j=1 £=1
where q ~s a polynomial and A
j£ are complex constants, q(O) + fj,£A j £ O. Hence we obtain
(7) (k 1,2,3, ... ),
where qk ~s the k-th term of q(z) and hj£k ~s the k-th Taylor-coefficient
z -£ of (I - - ) .
a.
J
Let us consider this Taylor-coefficient separately. By the binomial ser~es
expansion we have
00
(8)
f
k=O
and hence, by repeated application of the addition property ~n Pascal's
triangle
(9) (1 __ a z)-£
Therefore, the k-th Taylor-coefficient of (I - ~)-£ ~s equal to
a.
(10) a. -k
J
4 -(k) -k+t k (t) (zk) (t) Recall that t o. j =:
tT
(z ) z = : -1 • where Z=
a. ex • J Jdenotes the value of the t-th derivative of the function 2 k taken at 2
a..
Hence, the contribution of the term A.
(1-J9,
Z -9,
to the value of the J
ro a..
J
desired expression equal to
9,-1
£-1 F(t) (_I )
A.
I
( t )Jil,
t=O t!a. t a..
J J
1_11
_1provided that < (4pq) 2 (cf .
(3» .
a.. J
3. We now apply our results to the well-known convolutional code over GF(2)
' h 1 ' 1 2 2
w~t generator po ynom~a s I + D + D and I + D • The function G = G
E to be taken for the construction of the union bound
for the first event error prob ab i l i ty P
E ~n [ 1 ] ~s equal to 5 1 I I 2 I 3 I 4 1 G E (z) Z :=
=-- -
T6z-8
2 - - z - - z + 32 1 - 2z 32 4 2So we find for the union bound for P E
_1 1 3
(l - 2p)(1 - 16pq) 2 -
3'2-
pFor the construction of the union bound for the bit error probability P
B in [I] we must take G
=
G B 1 z5 I 3 1 2 1 3 GB(z) := ----2:0- =='8
+ T6 z + 4"z +4"Z -
5 ~~- + - - ---:;,-I - 2z 32 (1 _ 2z) (l - 2z)Hence we find as union bound for P B PB < ]36PI+-t-P2+-t-P3 - }2 F(2) + 1 [F(2) + 2F'(2)] = 5 7 3 2 1 3 I 32 + T6P+ 4"P - Zp -
32
(1-2p) -96pq)(I-16pq) 3 25
-1
These bounds hold for pq <
16
4. The first attempt to find an upper bound for the un10n bounds for P
E and P
B was made by Viterbi [IJ. Van de Meeberg [2J improved upper
bound using only the fact that P
Zk
=
P2k- 1 (k=
1,2,3, ... ). Experimen-tal measurements also have been made on PH for the special code with2 2
generators 1 + D + D and 1 + D • It turns out numerically that the union bound on PH for this code is closer to Van de Meeberg's bound than to the experimental results. So it seems to be useful now to study the philosophy behind Viterbi's union bound in order to obtain better bounds.
[IJ A.J. Viterbi. Convolutional codes and their performance 1n
[2J
communications systems.
IEEE Trans. Comm. Techn. COM-19 (1971), 751-772.
L. van de Meeberg. A ghtened upper bound on the error probability
of binary convolutional codes with Viterbi decoding. IEEE Trans. Inf. Theory IT-20 (1974), 389-391.