ECS 452: Digital Communication Systems 2013/1

HW 1 — Due: July 24
Lecturer: Prapun Suksompong, Ph.D.

Instructions

(a) ONE part of a question will be graded (5 pt). Of course, you do not know which part
will be selected; so you should work on all of them.

(b) It is important that you try to solve all problems. (5 pt)

(c) Late submission will be heavily penalized.

(d) Write down all the steps that you have done to obtain your answers. You may not get
full credit even when your answer is correct without showing how you get your answer.

Problem 1. Consider the two-path channels in which the receive signal is given by

y(t) = β1 x(t − ∆t1 ) + β2 x(t − ∆t2 ).

Four different cases are considered.

(a) Small |∆t1 − ∆t2 | and |β1 |  |β2 |

(b) Large |∆t1 − ∆t2 | and |β1 |  |β2 |

(c) Small |∆t1 − ∆t2 | and |β1 | ≈ |β2 |

(d) Large |∆t1 − ∆t2 | and |β1 | ≈ |β2 |

Figure 1.1 shows four plots of normalized1 |X(f )| (dotted black line) and normalized
|Y (f )| (blue line) in [dB]. Match the four graphs (i-iv) to the four cases (a-d).

Problem 2. Consider four vectors

       
1+j 1 1 −1
(1) (2) (3) (4)
v = 1 − j , v = −1 , v =
     1 , and v = −1 .
 
0 0 −1 −j
1
The function is normalized so that the maximum point is 1 dB.

1-1
 ECS 452 HW 1 — Due: July 24 2013/1

(i) (ii)

0 0

-10 -10
[dB]

[dB]
-20 -20

-30 -30
-1.5 -1 -0.5 0 0.5 1 1.5 -1.5 -1 -0.5 0 0.5 1 1.5
f f
(iii) (iv)

0 0

-10 -10
[dB]

[dB]
-20 -20

-30 -30
-1.5 -1 -0.5 0 0.5 1 1.5 -1.5 -1 -0.5 0 0.5 1 1.5
f f

Figure 1.1: Frequency selectivity in the receive spectra (blue line) for two-path channels.

(a) Use the Gram-Schmidt orthogonalization procedure (GSOP) (where the vectors are
applied in the order given) to the orthonormal vectors e(1) , e(2) , . . . that can be used
to represent v (1) , v (2) , v (3) , and v (4) .

(b) Use the orthonormal

 vectors from the previous part to construct the matrix E =
 (1)
e e · · · . Find the matric C such that V = EC where V = v v (2) v (3) v (4) .
(1) (2)

Hint: Implement GSOP in MATLAB.

78 Problem 3. ConsiderOptimum
the two signals
receivers1for
(t) binary
and s2 (t) shown
data in Figure 1.2. Note that V and
transmission
Tb are some positive constants. Your answers should be given in terms of them.

s1(t) s2(t)

V V

Tb
t t
0 0 Tb

−V
(a)
Figure 1.2: Signal set for Question 3
φ 1(t) φ 2(t)
(a) Find the energy in each signal.
1 Tb 1 Tb
1-2
Tb
t t
0 0 Tb

−1 Tb
 To obtain the geometrical picture, consider the case when α = 14 Tb . For this value of α
√
one has ρ = − 12 . As before φ1 (t) = s1 (t)/ E, whereas the second orthonormal function
is given by

2 1
ECS 452 φ2 (t) =
HW√1 — √ Due:s2 (t)
July+ 24s1 (t) . (5.38)
2013/1
3V Tb 2
The two orthonormal basis functions are plotted in Figure 5.8. The geometrical representa-
tion(b)
of sUse the Gram-Schmidt orthogonalization
1 (t) and s2 (t) is given in Figure 5.9. Noteprocedure (GSOP) (where
that the coefficients therepresentation
for the signals are
applied in the1 √ √ √
order given) to find two orthonormal functions φ (t) and φ (t) that can
= − = 3 2 + s2 = E, the 1 2
of s2 (t) are s E
√ used to represent
be 21 2 and s 22
s1 (t) and s22 (t).E. Since s 21 22 signal s2 (t) is at a dis-
tance of E from the origin. In general as α varies from 0 to Tb , the function φ2 (t) changes.
(c) Find the two vectors that represent the two waveforms s1 (t) and√s2 (t) in the new
However, for each specific φ2 (t) the signal s2 (t) is always at distance E from the origin.
(signal) space based on the orthonormal basis found in the previous part. Draw the
The locus of s2 (t) is also
corresponding plotted in Figure 5.9. Note that as α increases, ρ increases and
constellation.
the distance between the two signals decreases. 
Problem 4. Consider the two signals s1 (t) and s2 (t) shown in Figure 1.3. Note that V , α
and Tb are some positive constants.

s1(t) s2(t)

V V

α Tb


t t
0 Tb 0

−V

Fig. 5.7 Signal set for Example 5.3. Figure 1.3: Signal set for Question 4

φ 1(t) in each signal.

(a) Find the energy φ 2(t)

(b) Use the Gram-Schmidt orthogonalization 3procedure

3Tb (GSOP) (where the signals are
applied in
1 the
Tb order given) to find two orthonormal functions φ1 (t) and φ2 (t) that can
be used to represent s1 (t) and s2 (t). 1
T b Tb


4
Tb t t
(c) Plot φ1 (t) and
0 φ2 (t) whenTα
b
= 4
. 0
− 1 3T
(d) Find the two vectors s(1) and s(2) that representb the two waveforms s1 (t) and s2 (t) in
the new (signal) space based on the orthonormal basis found in the previous part.
Fig. 5.8 Orthonormal functions for Example 5.3.
(e) Draw the corresponding constellation when α = T4b .
k
(f) Draw s(2) when α = T
10 b
where k = 1, 2, . . . , 9.

1-3
HW1 Q1 Two-Path Channels
Monday, July 08, 2013 1:53 PM

ECS452 HW1 Page 1

HW1 Q2 GSOP for Complex-Valued Vectors
Monday, July 08, 2013 1:48 PM

ECS452 HW1 Page 2

ECS452 HW1 Page 3
HW1 Q3 Signal Space and Constellation
Monday, July 08, 2013 10:39 AM

ECS452 HW1 Page 4

HW1 Q4 Signal Space and Constellation
Monday, July 08, 2013 11:17 AM

ECS452 HW1 Page 5

ECS452 HW1 Page 6
ECS452 HW1 Page 7
ECS 452: Digital Communication Systems 2013/1
HW 2 — Due: Not Due
Lecturer: Prapun Suksompong, Ph.D.

Problem 1. In a binary antipodal signaling scheme, the message S is randomly selected

from the alphabet set S = {3, −3} with p1 = P [S = −3] = 0.3 and p2 = P [S = 3] = 0.7.
The message is corrupted by an independent additive exponential noise N whose pdf is
 1 −n/2
2
e , n ≥ 0,
fN (n) =
0, otherwise.

(a) Find the MAP detector ŝMAP (r).

(b) Indicate the decision regions of the MAP detector in part (a).

(c) Consider a detector of the form


3, r > τ,
ŝ (r) =
−3, r ≤ τ

for some threshold τ . Find and then plot the probability of (symbol detection) error
for this detector as a function of τ . Hint: The plots from actual simulation are shown
in class. The same plots are shown in Figure 2.1.

(d) Evaluate the error probability of the MAP detector.

(e) Evaluate the error probability of the ML detector.

Problem 2. Repeat parts (a)-(d) of Question 1 but now the noise is uniform on [−4, 4].
Problem 3. In a ternary signaling scheme, the message S is randomly selected from the
alphabet set S = {−1, 1, 4} with p1 = P [S = −1] = 0.3 = p2 = P [S = 1] and p3 =
P [S = 4] = 0.4. The message is corrupted by an independent additive Gaussian noise
N ∼ N (0, 2).
(a) Find the average signal energy1 Es .
Note that X
Es = pi |si |2 .
i

1
Same as “average symbol energy” or “average energy per symbol” or “average energy per signal”

2-1
ECS 452 HW 2 — Due: Not Due 2013/1

0.7

0.6

0.5

0.4
P(E)

0.3

0.2

0.1

0
-10 -8 -6 -4 -2 0 2 4 6 8 10


8
Figure 2.1: P (E) for Exponential Noise in Question 1

(b) Find the MAP detector ŝMAP (r).

(c) Indicate the decision regions of the MAP detector in part (b).

(d) Evaluate the error probability of the MAP detector.

Problem 4. In a ternary signaling scheme, the message S is randomly selected from the
alphabet set S = {−1, 1, 4} with p1 = P [S = −1] = 0.41, p2 = P [S = 1] = 0.08 and
p3 = P [S = 4] = 0.51. The message is corrupted by an independent additive Gaussian noise
N ∼ N (0, 2).

(a) Find the average signal energy Es .

(b) If the MAP detector is used, find P (E|S = 1); that is, find the probability of (decoding)
error given that S = 1 was transmitted.

Problem 5. In a standard quaternary  signaling scheme, the message S is equiprobably

3d d d 3d

selected from the alphabet set S = − 2 , − 2 , 2 , 2 . The message is corrupted by an
independent additive exponential noise N whose pdf is
 −λ
λe , n ≥ 0,
fN (n) =
0, otherwise.

(a) Find the average symbol energy.

2-2
ECS 452 HW 2 — Due: Not Due 2013/1

(b) Find the average energy per bit.

(c) Find the MAP detector ŝMAP (r).

(d) Evaluate the error probability of the MAP detector.

Eb
(e) Let λ = σ1 . (This is to set Var N = σ 2 as in the case for Gaussian noise.) Plot σ2
vs.
probability of error P (E). Consider Eσ2b from -30 to 10 dB.

2-3
HW2 Q1: 1-D MAP Detector and Exponential Noise
Monday, July 15, 2013 10:58 AM

ECS452 HW2 Page 1

ECS452 HW2 Page 2
ECS452 HW2 Page 3
HW2 Q2: 1-D MAP Detector and Uniform Noise
Monday, July 15, 2013 1:33 PM

ECS452 HW2 Page 4

ECS452 HW2 Page 5
ECS452 HW2 Page 6
ECS452 HW2 Page 7
HW2 Q3: 1-D Multi-Level MAP Detector and Gaussian Noise
Wednesday, July 24, 2013 2:31 PM

ECS452 HW2 Page 8

ECS452 HW2 Page 9
HW2 Q4: 1-D Multi-Level MAP Detector and Gaussian Noise
Wednesday, July 24, 2013 3:13 PM

ECS452 HW2 Page 10

HW2 Q5: 1-D Standard Multi-Level MAP Detector and Expo Noise
Wednesday, July 24, 2013 3:35 PM

ECS452 HW2 Page 11

ECS452 HW2 Page 12