Scribd
Upload
8 views3 pages

HW01

This document outlines Homework #1 for COM 512000 Communication Theory, due on September 27, 2006. It includes various problems related to Gaussian random variables, complex random processes, power spectral density, filtering, and sampling, with specific tasks requiring detailed derivations for full credit. The homework consists of multiple sections, each focusing on different aspects of communication theory and random processes.

Uploaded by

looper-handier.5g
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
8 views3 pages

HW01

This document outlines Homework #1 for COM 512000 Communication Theory, due on September 27, 2006. It includes various problems related to Gaussian random variables, complex random processes, power spectral density, filtering, and sampling, with specific tasks requiring detailed derivations for full credit. The homework consists of multiple sections, each focusing on different aspects of communication theory and random processes.

Uploaded by

looper-handier.5g
Copyright
© © All Rights Reserved
We take content rights seriously. If you suspect this is your content, claim it here.
Available Formats
Download as PDF, TXT or read online on Scribd
You are on page 1/ 3

COM 512000 Communication Theory

Homework #1
(Due September 27, 2006 BEFORE class)

Note: Detailed derivations are required to obtain a full score for each problem. (Total 120%)

1. (16%) [Gaussian random variable and the Chernoff Bound]


(a) Show that the probability of the event X > x is bounded by

Pr[X > x] ≤ e−sx ΦX (s),

for any real-valued s ≥ 0, where ΦX (s) is the moment generating function of X.


(b) Find a similar bound
FX (x) ≤ esx ΦX (−s)

for s ≥ 0.
(c) Show that the moment generating function of a Gaussian random variable with mean µ
and variance σ 2 is
log ΦX (s) = s · µX + s2 · σX
2
/2.
2 /2
(d) Show that Q(x) ≤ e−x for x > 0.

2. (16%) [Complex Random Processes] Let Z(t) = X(t)+ jY (t) be a complex random
process.
(a) Assume that Z(t) = Z = X + jY , where X, Y are random variables. Show that Z(t) a
strictly stationary random process? Is Z(t) ergodic in the mean?
(b) Let Zi = Xi + jYi , for i = 1, · · · , n, be complex Gaussian random variables with mean
Pn j2πf0 t .
zero and let V (t) = i=1 Zi e Show that V (t) is wide-sense stationary.
(c) Following from (b), give the conditions for which the real part of V (t), i.e., ℜ{V (t)}, is
wide-sense stationary. Under these conditions, is V (t) stationary in the strict sense?
(d) Independent from the questions above, what are the conditions that must be satisfied
in order to make W (t) = X(t) cos 2πf t + Y (t) sin 2πf t wide-sense stationary.

3. (5%+5%+10%) [Power Spectral Density] Let X(t) be a real-valued Wide-Sense


Stationary (WSS) random process with the autocorrelation function RXX (τ ) = E[X(t +
τ )X(t)] and the power density spectrum SX (f ).
(a) Show that SX (f ) = SX (−f ).
(b) Show that SX (f ) is real.
(c) Determine which of the following functions can and cannot be a valid power density
spectrum. For those that are not, explain why.

1
f2 f2 f4 cos(3f )
(i) f 6 +3f 2 +3
; (ii) exp[−(f − 1)2 ]; (iii) f 4 +1
− δ(f ); (iv) 1+f 2 +jf 6
; (v) 1+f 2
where δ(f ) is the Dirac delta function.

4. (16%) [PSD and Filtering] (a) Find the autocorrelation function of the random
process W (t) that has the power spectrum
8
SW (f ) = .
[9 + (2πf )2 ]2
(b) Suppose we want to construct a Gaussian noise process with the spectrum given in (a)
by passing the Gaussian white noise process N (t) through a linear time invariant filter h(t).
Let N (t) be zero-mean and has variance N0 . Find the filter h(t) that gives us the desired
noise process at the output.

5. (3%+4%+10%+10%) [Sampling] Consider a WSS Gaussian random process X(t)


 2
sin πW τ
with 0-mean and autocorrelation function equal to RXX (τ ) = πτ .
(a) What is the minimum rate for which we need to sample in order to reconstruct the
random signal X(t)?
(b) By sampling at the minimum rate, what is the discrete time autocorrelation function
of the samples? Are they WSS?;
(c) Let {X[m] = X(mT )} be the samples taken from X(t), where 1/T is the minimum
sampling rate. We obtain a reconstruction of the signal
∞
X
Y (t) = X[k]g(t − kT ).
k=−∞

Let G(f ) be the Fourier transform of g(t). Derive the average power density spectrum of
Y (t).
(d) Show that
n o
E |X(t) − Y (t)|2 = 0
sin(πt/T )
for g(t) = πt/T where 1/T is the minimum sampling rate. (Do not worry about the
interchange between the infinite summation and the expectation.)

6. (25%) [PSD and Filtering] Let s(t) be a real-valued deterministic signal (with
Fourier transform S(f )) transmitted through an Additive Noise Channel and let n(t) be
an additive noise modeled as a stationary Gaussian Random Process with mean zero and
N0
autocorrelation πnn (τ ) = 2 δ(τ ) where δ() represents the delta function.
As shown in Fig. 1, the signal that arrives at the receiver is modeled as

r(t) = s(t) + n(t).

At the receiver, we pass the received signal r(t) through a Linear Time-Invariant filter with
a frequency transfer function represented by H(f ). The sample of the output z(t) at time

2
n(t)
T
s(t) r(t) z(t) z(T )
+ LTI

Figure 1: Matched Filtering

T is then used for detection.


(a) (6%) Derive the power density spectrum of the random process n(t). What is the
instantaneous power of the noise component in z(T ), i.e., E[|b(T )|2 ], expressed with H(f )?
(b(T ) is the noise component in z(T ))
(b) (7%) What is the instantaneous power of the deterministic signal component in z(T ),
i.e., |a(T )|2 where a(T ) is the signal component in z(T )? (Express it with S(f ) and H(f ).)
(c) (12%) Let us define
S |a(T )|2
=
N E[|b(T )|2 ]
as the signal-to-noise ratio of z(T ). Derive the value of H(f ) that maximizes the signal-to-
noise ratio. What is the expression of h(t) in terms of s(t)?
(Hint: Schwarz’s Inequality
Z ∞ 2 Z ∞ Z ∞
g1 (x)g2 (x)dx ≤ |g1 (x)|2 dx |g2 (x)|2 dx
−∞ −∞ −∞

The equality holds when g1 (x) = kg2∗ (x).)

You might also like

pFad - Phonifier reborn

Pfad - The Proxy pFad of © 2024 Garber Painting. All rights reserved.

Note: This service is not intended for secure transactions such as banking, social media, email, or purchasing. Use at your own risk. We assume no liability whatsoever for broken pages.


Alternative Proxies:

Alternative Proxy

pFad Proxy

pFad v3 Proxy

pFad v4 Proxy