Random Signal 

Analysis
ECE 5610/4610 Lecture Notes

r�0 r = 0 r�0
y y y

f � x� y � f � x� y � f � x� y �

x x x

x � t� � 0 � y � t� � i �� i = 0� 1� �

h LP � t � t
t

© 1990 – 2004
Mark A. Wickert
�
Chapter 1
Course Introduction/Overview
Contents

1.1 Introduction to Random Signals . . . . . . . . . . . . . . . . . . . . . . . . . 1-3

1.2 Mathematical Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-4
1.3 Engineering Applications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-6
1.4 Random Signals in Practice . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-7
1.5 Course Perspective in the Comm/DSP Area of ECE . . . . . . . . . . . . . . 1-22
1.6 What is this course about? . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-23
1.7 The Role of Computer Analysis/Simulation Tools . . . . . . . . . . . . . . . . 1-24
1.8 Instructor Policies . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-25
1.9 Course Syllabus . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1-26

1-1
CHAPTER 1. COURSE INTRODUCTION/OVERVIEW

1-2 ECE 5610/4610 Random Signals

 1.1. INTRODUCTION TO RANDOM SIGNALS

1.1 Introduction to Random Signals

• Mathematical models
• Random signals in practice
• Course perspective
• What is this course about?
• The computer simulation project
• Instructor policies
• Course syllabus

ECE 5610/4610 Random Signals 1-3

CHAPTER 1. COURSE INTRODUCTION/OVERVIEW

1.2 Mathematical Models

• Mathematical models serve as tools in the analysis and design

of complex systems

• A mathematical model is used to represent, in an approximate

way, a physical process or system where measurable quantities
are involved

• Typically a computer program is written to evaluate the mathe-

matical model of the system and plot performance curves

– The model can more rapidly answer questions about system

performance than building expensive hardware prototypes

• Mathematical models may be developed with differing degrees

of fidelity

• A system prototype is ultimately needed, but a computer simu-

lation model may be the first step in this process

• A computer simulation model tries to accurately represent all

relevant aspects of the system under study

• Digital signal processing (DSP) often plays an important role in

the implementation of the simulation model

• If the system being simulated is to be DSP based itself, the simu-

lation model may share code with the actual hardware prototype

• The mathematical model may employ both deterministic and

random signal models
1-4 ECE 5610/4610 Random Signals
 1.2. MATHEMATICAL MODELS

Formulate
Hypothesis

Define Experiment to
Test Hypothesis

Physical Modify
or Simulation of Model
Process/System

Observations Predictions

Sufficient No
Agreement?

No All Aspects
of Interest
Investigated?

Stop

The Mathematical Modeling Process1

1 Alberto Leon-Garcia, Probability and Random Processes for Electrical Engineering, Addison-Wesley, Reading,
MA, 1989

ECE 5610/4610 Random Signals 1-5

CHAPTER 1. COURSE INTRODUCTION/OVERVIEW

1.3 Engineering Applications

Communications, Computer networks, Decision theory and decision

making, Estimation and filtering, Information processing, Power en-
gineering, Quality control, Reliability, Signal detection, Signal and
data processing, Stochastic systems, and others.

Relation to Other Subjects2

Estimation
and Filtering Information
Signal Theory
Processing

Communication
Reliability & Wireless
Random Signals
and Systems
Linear
Systems
Others
Probability
Random
Variables
Statistics
Decision
Theory
Game Mathematics
Theory

2 X. Rong Li, Probability, Random Signals, and Statistics, CRC Press, Boca Raton, FL, 1999

1-6 ECE 5610/4610 Random Signals

 1.4. RANDOM SIGNALS IN PRACTICE

1.4 Random Signals in Practice

• A typical application of random signals concepts involves one

or more of the following:
– Probability
– Random variables
– Random (stochastic) processes

Example 1.1: Modeling with Probability

• Consider a digital communication system with a binary symmet-

ric channel and a coder and decoder
Input Output
1–ε
0 0
Binary ε
Channel ε
Model
1 1
1–ε
ε = Error Probability

Binary Received
Information Binary Information
Coder Channel Decoder

Communication System with Error Control

A data link with error correction

• The channel introduces bit errors with probability Pe �bit) = �

• A simple code scheme to combat channel errors is to repetition
code the input bits by say sending each bit three times
ECE 5610/4610 Random Signals 1-7
CHAPTER 1. COURSE INTRODUCTION/OVERVIEW

• The decoder then decides which bit was sent by using a majority
vote decision rule
• The system can tolerate one channel bit error without the de-
coder making an error
• A symbol error occurs when either two or three channel bit er-
rors occur
• The probability of a symbol error is given by
Pe �symbol) = P�2 bit errors) � P�3 bit errors)
• Assuming bit errors are statistically independent we can write
P�2 bit errors) = � · � · �1 − �) � � · �1 − �) · �
� �1 − �) · � · � = 3� 2�1 − �)
P�3 errors) = � · � · � = � 3
• The symbol error probability is thus
Pe �symbol) = 3� 2 − 2� 3
• Suppose Pe �bit) = � = 10−3, then Pe �symbol) = 2.998 × 10−6
• The error probability is reduced by three orders of magnitude,
but the coding reduces the throughput by a factor of three

Example 1.2: Modeling with Random Variables

• Here we assume that a voltage x is measured as being only noise,

or noise plus signal
�
n, only noise
x=
v � n, noise + signal

1-8 ECE 5610/4610 Random Signals

 1.4. RANDOM SIGNALS IN PRACTICE

• We model x as a random variable with a probability density func-

tion dependent upon which hypothesis is present

fx � x n ) fx � x v + n )
Area = 1
vT

x
–1 0 v–1 1 v v+1

Conditional density function on x

• We decide that the hypothesis signal is present if x > vT , where

vT is the so-called decision threshold
• The probability of detection is given by
� ∞
PD = P�x > vT |v � n) = f x �x|v � n) d x
vT

fx � x v + n )
PD = P � x ≥ vT v + n )
vT

x
–1 0 v–1 1 v v+1

Area corresponding to PD

Example 1.3: Modeling with Random Processes

• Consider a random or stochastic process of the form

x�t) = A cos�2π f c t � θ ) � n�t)
which is a sinusoidal carrier plus noise
ECE 5610/4610 Random Signals 1-9
CHAPTER 1. COURSE INTRODUCTION/OVERVIEW

• In this example the carrier phase θ is modeled as a random vari-

able and n�t) is modeled as an independent stationary random
process
• We may be interested in how to recover the sinusoidal carrier
from the noisy signal x�t)
• The power spectral density of a random process allows us to see
the spectral content of a signal
• The power spectral density of a wide sense stationary random
process x�t) is given by the Fourier transform of the autocorre-
lation function
• In this case the power spectrum is
A2
Sx x � f ) = [δ� f − f c ) � δ� f � f c )] � Snn � f )
4
where Snn � f ) is the power spectrum of the noise alone
S nn � f )
S xx � f )

f
–fc 0 fc

Power spectral density of x�t)

• To recover just the carrier from x�t) we may pass x�t) through
a filter

1-10 ECE 5610/4610 Random Signals

 1.4. RANDOM SIGNALS IN PRACTICE

A cos � 2πf c t + θ ) Filter y�t)

n�t)

Signal processing x�t) to recover just the carrier signal

S xx � f )
Highpass or
Bandpass Filter

f
–fc 0 fc
S yy � f )

f
–fc 0 fc

Filtering x�t) to obtain y�t) with spectrum Syy � f )

Example 1.4: Multiple User Communication Environments

• Code Division Multiple Access (CDMA) is utilized in one of the

second generation mobile communications systems, i.e., IS-95
• To model the system performance of this system we can use
random signals
• A simplified block diagram of an equivalent baseband system is
shown below
ECE 5610/4610 Random Signals 1-11
CHAPTER 1. COURSE INTRODUCTION/OVERVIEW

d1 � t ) WGN: n � t )
Delay, τ 1

c1 � t )
d2 � t )
Delay, τ 2

y� t)
c2 � t )

dK � t ) . . .
Delay, τ K
Tb dˆ 1 � t )
cK� t )
�0 � . ) dt
c1 � t – τ1 )

Detection of user 1 in a K users CDMA system

• The composite received signal can be written as

�
y�t) = P1/2 d1�t − τ1)c1�t − τ1)
K
� �
� Pk /2 dk �t − τk )ck �t − τk ) � n�t)
k=2
where Pk and τk denote the signal power and propagation delay,
respectively, for the k-th user, and n�t) is white Gaussian noise
• The signals ck �t) are the unique spreading codes associated with
each user and the signals dk �t) are the user data
• Here we assume c�t) and d�t) take on values of ±1 over the bit
interval
• Ideally, we choose the spreading codes to be mutually orthogo-
nal, e.g., Walsh codes are used in IS-95
1-12 ECE 5610/4610 Random Signals
 1.4. RANDOM SIGNALS IN PRACTICE

• Multipath propagation, not modeled here, will prevent perfect

orthogonality from being maintained at the receiver
• Assuming a local despreading code of the form c1�t − τ1) we
have perfect synchronization, and we can write the integrator
output for the data bit on 0 ≤ t ≤ Tb as
K
�
� �
Y = P1/2 d1�0)Tb � Pk /2 Tb dk �0)ρ1k � N g
k=2
where the first term is the desired signal, the second term con-
stitutes multiple access noise, and the third term is a Gaussian
random variable due to the AWGN channel noise
• The multiple access noise is controlled in part by the aperiodic
correlation coefficient from
� τuser 1 to user k
dk �−1) k
ρ1k = c1�t)ck �t � Tb − τk ) dt
dk �0) 0
� Tb
� c1�t)ck �t − τk ) dt
τk

• The decision rule applied to Y is to declare a +1 is sent if Y > 0

and a -1 is sent if Y < 0
• The exact statistics associated with the random variable Y are
quite complex
• The multiple access noise can be approximated as a Gaussian
random variable and hence result in a rather simple form for the
bit error probability (BEP)
• In a paper by Pursley3 it is shown that
�√ �
BEP � Q SNR
3 M.B. Pursley, “Performance Evaluation of Phase-Coded Spread-Spectrum Multiple-Access Communications”,
IEEE Trans. Commun., Vol. COM-25, pp. 800–803, Aug. 1977.

ECE 5610/4610 Random Signals 1-13

CHAPTER 1. COURSE INTRODUCTION/OVERVIEW

where
N0 −1
��
K −1
SNR = �
3N 2E b
K is the number of users and N is the number of spreading code
chips per bit, i.e., the processing gain; note here we assume per-
fect power control so all of the received signal powers are equal

• The term E b /N0 is the ratio of bit energy to noise power spectral
density, usually given in dB as 10 log10�E b /N0)

• The function Q�) is the Gaussian Q-function which is the area

under the tail of a zero mean unit variance Gaussian random
variable � ∞
1 2
Q�x) = √ e−u /2 du
2π x
• A family of BEP performance curves is shown below
CDMA Bit Error Probability
0
N = 64, Like IS-95
−1

−2
K = 30 users
log 10 [ BEP ]

K = 20 users
−3

−4
K = 10 users
−5
K = 1 K = 5 users
−6
0 5 10 15 20 25 30
E b ⁄ N 0  dB

1-14 ECE 5610/4610 Random Signals

 1.4. RANDOM SIGNALS IN PRACTICE

Example 1.5: Simulation using Random Processes

• The second generation wireless system Global System for Mo-

bile Communications (GSM), uses the Gaussian minimum shift-
keying (GMSK) modulation scheme
� � ∞
� �
xc �t) = 2Pc cos 2π f 0t � 2π f d an g�t − nTb )
n=−∞

where
� � � � ��
1 2 t 1
g�t) = erf − π BTb −
2 ln 2 Tb 2
�� � � ��
2 t 1
� erf π BTb �
ln 2 Tb 2
• The GMSK shaping factor is BTb = 0.3 and the bit rate is Rb =
1/Tb = 270.833 kbps
• We can model the baseband GSM signal as a complex random
process
• Suppose we would like to obtain the fraction of GSM signal
power contained in an RF bandwidth of B Hz centered about
the carrier frequency
• There is no closed form expression for the power spectrum of a
GMSK signal
• A simulation constructed in MATLAB can be used to produce a
complex baseband version of the GSM signal

ECE 5610/4610 Random Signals 1-15

CHAPTER 1. COURSE INTRODUCTION/OVERVIEW

>> [x,data] = gmsk(0.3, 10000, 6, 16, 1);

>> [Px,F] = psd(x,1024,16);
>> [Pbb,Fbb] = bb_spec(Px,F,16);
>> plot(Fbb,10*log10(Pbb));
>> axis([-400 400 -60 20]);

20

10

-10
P fraction
-20

-30

-40 B

-50

-60
-400 -300 -200 -100 0 100 200 300 400

Estimated baseband GSM power spectra

• Using averaged periodogram spectral estimation we can esti-

mate SGMSK� f ) and then find the fractional power in any RF
bandwidth, B, centered on the carrier
� B/2
−B/2 SGMSK� f ) d f
Pfraction = � ∞
−∞ SGMSK� f ) d f

– The integrals become finite sums in the MATLAB calcula-

tion
1-16 ECE 5610/4610 Random Signals
 1.4. RANDOM SIGNALS IN PRACTICE

GSM Power Containment vs. RF Bandwidth
1

Fractional Power 0.8

0.6
B = 200 kHz ↔ 95.6%
0.4 B = 100 kHz ↔ 67.8%
B = 50 kHz ↔ 38.0%
0.2

0
0 50 100 150 200 250 300
RF Bandwidth in kHz

Fractional GSM signal power in a centered B Hz RF bandwidth

• An expected result is that most of the signal power (95%) is con-

tained in a 200 kHz bandwidth, since the GSM channel spacing
is 200 kHz

Example 1.6: Separate Queues vs A Common Queue

A well known queuing theory result4 is that multiple servers, with
a common queue for all servers, gives better performance than mul-
tiple servers each having their own queue. A chapter on queuing
theory is contained near the end of the course text. It is interesting to
see probability theory in action modeling a scenario we all deal with
in our lives.
4 Mike Tanner, Practical Queuing Analysis, The IBM-McGraw-Hill Series, New York, 1995.

ECE 5610/4610 Random Signals 1-17

CHAPTER 1. COURSE INTRODUCTION/OVERVIEW

Customers Servers
One
Queue 1
Random (waiting line)
Arrivals at λ 2 Departing
per unit time
Customers
(exponentially

...
distributed) Rate
λ m

T s =  Average Service
Time

Separate
Customers Servers
Queues

1
Random
Arrivals at λ 2 Departing
per unit time Customers
(exponentially
...

distributed)
m
Rate
Assume customers T s =  Average Service
λ⁄m Time
randomly pick queues
One long service-time customer 
forces those behind into a long wait

Common queue versus separate queues for multiple servers

Common Queue Analysis

• The number of servers is defined to be m, the mean customer
arrival rate is λ per unit of time, and the mean customer service
time is Ts units of time
• In the single queue case we let u = λTs = traffic intensity
• Let ρ = u/m = server utilization
• For stability we must have u < m and ρ < 1
1-18 ECE 5610/4610 Random Signals
 1.4. RANDOM SIGNALS IN PRACTICE

• As a customer we are usually interested in the average time in

the queue, which is defined as the waiting time plus the service
time (Tanner)
E c �m, u)Ts
TQCQ = Tw � Ts = � Ts
� m�1 − ρ) �
E c �m, u) � m�1 − ρ) Ts
=
m�1 − ρ)
where
u m /m!
E c �m, u) =
u m /m! � �1 − ρ) m−1
� k
k=0 u /k!

is known as the Erlang-C formula

• To keep this problem in terms of normalized time units, we will
plot TQ /Ts versus the traffic intensity u = λTs
• The normalized queuing time is
TQCQ E c �m, u) � �m − u) E c �m, u)
= = �1
Ts m−u m−u
Separate Queue Analysis
• Since the customers randomly choose a queue, arrival rate into
each queue is just λ/m
• The server utilization is ρ = �λ/m) · Ts which is the same as the
single queue case
• The average queuing time is (Tanner)
Ts Ts mTs
TQSQ = = =
1−ρ 1 − λT
m
s m − λTs

ECE 5610/4610 Random Signals 1-19

CHAPTER 1. COURSE INTRODUCTION/OVERVIEW

• The normalized queuing time is

TQSQ m m
= =
Ts m − λTs m−u

• Create the Erlang-C function in MATLAB:

function Ec = erlang_c(m,u);
� Ec = erlangc(m,u)
�
� The Erlang-C formula
�
� Mark Wickert 2001

s = zeros(size(u));
for k=0,
s = s + u.ˆk/factorial(k);
end

Ec = (u.ˆm)/factorial(m)./(u.ˆm/factorial(m) + (1 - u/m).*s);

• We will plot TQ /Ts versus u = λTs for m = 2 and 4

>> � m = 2 case
>> u = 0:.05:1.9;
>> m = 2;
>> Tqsq = m./(m - u);
>> Tqcq = erlang_c(m,u)./(m-u) + 1;
>> � m = 4 case
>> u = 0:.05:3.9;
>> m = 4;
>> Tqsq = m./(m - u);
>> Tqcq = erlang_c(m,u)./(m-u) + 1;

• The plots are shown below:

1-20 ECE 5610/4610 Random Signals
 1.4. RANDOM SIGNALS IN PRACTICE

7
m = 2 Separate
6

5
TQ
------- 4
Ts
3
Common
2

0
0 0.2 0.4 0.6 0.8 1 1.2 1.4 1.6 1.8 2
Traffic Intensity  = λT s

Queuing time of common queue and separate queues for m = 2 servers

8

7
m = 4
6 Separate
5
TQ
------- 4
Ts
3

2 Common
1

0
0 0.5 1 1.5 2 2.5 3 3.5 4
Traffic Intensity  = λT s

Queuing time of common queue and separate queues for m = 4 servers

ECE 5610/4610 Random Signals 1-21