You are on page 1of 3

Optical Communication Theory and Techniques

Part I: Communication Theory and Digital Transmission


September 11, 2013

1. Given the signals in the figure below:

s1 (t) s2 (t) s3 (t) s4 (t)


1 1 1

0 3 t 0 2 3 t 0 2 t 0 1 2 t
−1 −1 −1

(a) Find a basis for the space generated by the signals by using the Gram-Schmidt orthogo-
nalization procedure.
(b) Determine the vector representation of the signals over this basis.

2. A binary digital communication system employs the signals

s1 (t) = 0
0≤t≤T
s2 (t) = A

for transmitting the information. This is called on-off signaling.

(a) Determine the optimum detector for an AWGN channel, assuming that the signals are
equally probable.
(b) Determine the probability of error as a function of Eb /N0 , Eb being the average energy per
bit and N0 the noise power spectral density.
(c) Tell how on-off signaling compares with antipodal signaling.
Solution:

1. The first element of the basis is simply the normalized version of s1 (t), i.e.,
p
ϕ1 (t) = s1 (t)/ E1 ϕ1 (t)

´3 1/ 3
where E1 = 0 |s1 (t)|2 dt = 3.
0 3 t

The second element is obtained by subtracting to s2 (t) its projection onto ϕ1 (t), i.e.,
ϕ2 (t) = θ2 (t)/kθ2 k
θ2 (t) = s2 (t) − (s2 , ϕ1 )ϕ1 (t)
´3 √
where (s2 , ϕ1 ) = 0 s2 (t)ϕ1 (t)dt = 1/ 3, such that
s2 (t) (s2 , ϕ1 )ϕ1(t) θ2 (t)
1
2/3
− 1/3 =
0 2 3 t 0 3 t 0 2 3 t
−1
−4/3
´3  2  2 √ √
kθ2 k2 = |θ2 (t)|2 dt = 2
2+ 4
= 83 , and hence ϕ2 (t) = θ2 (t)/ 8/3 = √3 θ (t)
0 3 3 2 2 2

ϕ2 (t)

1/ 6
0 2 3 t

−2/ 6

In order to obtain the third element, we compute the projection of s3 (t) onto the space of
{ϕ1 (t),
√ ϕ2 (t)}, which is given
√ by (s3 , ϕ1 )ϕ1 (t) + (s3 , ϕ2 )ϕ2 (t). It can be easily seen that (s3 , ϕ1 ) =
−2/ 3 and (s3 , ϕ2 ) = −2/ 6, such that
(s3 , ϕ1 )ϕ1 (t) (s3 , ϕ2 )ϕ2(t) s3 (t)
2/3
0 3 + 0 2 3 =
t t 0 2 t
−1/3
−2/3
−1
 
and hence θ3 (t) = s3 (t) − (s3 , ϕ1 )ϕ1 (t) + (s3 , ϕ2 )ϕ2 (t) = 0, meaning that s3 (t) belongs to the
space generated by {ϕ1 (t), ϕ2 (t)} and we can simply discard it.
It can be easily verified that s4 (t) is orthogonal to both ϕ1 (t) and ϕ2 (t), such that the third element
of the sought basis is simply the normalized version of s4 (t). In conclusion the function of the
basis are
ϕ1 (t) ϕ2 (t) ϕ3√(t)
√ 1/ 2
1/ 3 √
1/ 6

0 3 t 0 2 3 t 0 1 2 t
√ √
−2/ 6 −1/ 2
2. (a) The optimum decision strategy is
(ˆ T )
m̂ = argmax r(t)si (t)dt + Ci
mi 0

where, as the signals are equally probable,



Ei 
0 ,
 i=1
Ci = − , Ei = 
 .
2  A2 T , i=2

As s1 (t) = 0, the decision strategy is equivalent to


ˆ T
E2 m1
r(t)s2 (t)dt − ≶0
0 2 m2

which, taking into account that s2 (t) = A and E2 = A2 T , may be rewritten as


ˆ
1 T A
choose m1 if r(t)dt < , m2 otherwise
T 0 2
and implemented either by using a correlator as follows:

1
RT z 
m 1 ,
 z < A/2
r(t) T 0
m̂ = 

A/2 m 2 , z > A/2

or a matched filter:

z 
m 1 ,
 z < A/2
r(t) h(t) m̂ = 

A/2 m 2 , z > A/2
t=T

where 

1/T,
 0≤t≤T
h(t) = 

0, otherwise
(b) In the binary case, the probability of error is given by
 d 
P(E ) = Q √
2N0

where d is the distance between the signals. In this case, d = E2 , and, as the signals are
equally probable, we have that Eb = (E1 + E2 )/2 = E2 /2. Thus,
r E 
b
P(E ) = Q
N0

(c) It is evident that on-off signaling is a case of orthogonal signaling, such that it has a penalty
of 3 dB with respect to antipodal signaling, as, in this last case,
r E 
b
P(E ) = Q 2
N0

You might also like