Upload
sitvijay
View
225
Download
0
Embed Size (px)
Citation preview
8/21/2019 EC 2351 Digital Communication
1/106
Digital CommunicationsChapter 2: Deterministic and Random Signal Analysis
Po-Ning Chen, Professor
Institute of Communications Engineering
National Chiao-Tung University, Taiwan
Digital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 1 / 106
8/21/2019 EC 2351 Digital Communication
2/106
2.1 Bandpass and lowpass signal representation
Digital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 2 / 106
8/21/2019 EC 2351 Digital Communication
3/106
2.1 Bandpass and lowpass signal representation
Definition (Bandpass signal)
A bandpass signal x (t ) is a real signal whose frequency content is located around central frequency f 0, i.e.X
(f
) = 0 for all
∣f
±f 0
∣ > W
X (f )
f 0f 0 −W f 0 +W −f 0 −f 0 +W −f 0 −W
f 0 may not be thecarrier frequency f c !
The spectrum of a bandpass signal is Hermitian symmetric ,
i.e., X (−f ) = X ∗(f ). (Why? Hint: Fourier transform.)Digital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 3 / 106
8/21/2019 EC 2351 Digital Communication
4/106
2.1 Bandpass and lowpass signal representation
Since the spectrum is Hermitian symmetric, we only need
to retain half of the spectrum X +(f ) = X (f )u −1(f )(named analytic signal or pre-envelope) in order toanalyze it,
where u −1(f ) = ⎧⎪⎪⎪⎪⎨⎪⎪⎪⎪⎩
1 f
> 0
1
2 f = 00 f
8/21/2019 EC 2351 Digital Communication
5/106
2.1 Bandpass and lowpass signal representation -
Baseband and bandpass signals
Definition (Baseband signal)A lowpass or baseband ( equivalent ) signal x (t ) is a complex signal (because it is not necessarily Hermitian symmetric!) whose spectrum is located around zero frequency, i.e.
X (f ) = 0 for all ∣f ∣ > W It is generally written as
x (t ) = x i (t ) + ı x q (t )where
x i
(t
) is called the in-phase signal
x q (t ) is called the quadrature signal Digital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 5 / 106
8/21/2019 EC 2351 Digital Communication
6/106
Baseband signal
Our goal is to relate x
(t
) to x
(t
) and vice versa
Digital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 6 / 106
8/21/2019 EC 2351 Digital Communication
7/106
From lowpass equivalent x (t ) to x (t )
Definition of bandwidth. The bandwidth of a signal is onehalf of the entire range of frequencies over which the spectrum
is (essentially) nonzero. Hence, W is the bandwidth in thelowpass signal we just defined, while 2W is the bandwidth of the bandpass signal by our definition.
Digital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 7 / 106
8/21/2019 EC 2351 Digital Communication
8/106
Analytic signal
Let’s start from the analytic signal x +
(t
).
x +(t ) = ∞−∞ X +(f )e ı2πft df
= ∞
−∞
X
(f
)u −1
(f
)e ı2πft df
= F −1 {X (f )u −1(f )} F −1 Inverse Fourier transform= F −1 {X (f )} ⋆ F −1 {u −1(f )}=
x
(t
) ⋆ 1
2δ
(t
) + ı
1
2πt
= 12x (t ) + ı 12 x̂ (t ),where x̂
(t
) = x
(t
) ⋆ 1πt
= ∫ ∞−∞
x (τ )π(t −τ )d τ is a real-valued signal.
Digital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 8 / 106
8/21/2019 EC 2351 Digital Communication
9/106
Appendix: Extended Fourier transform
F −1 {2u −1(f )} = F −1 {1 + sgn(f )}
= F −1
{1
} + F −1
{sgn
(f
)} = δ
(t
) + ı
1
πt
Since ∫ ∞−∞ ∣sgn(f )∣ = ∞, the inverse Fourier transform of sgn(f )does not exist in the standard sense! We therefore have toderive its inverse Fourier transform in the extended sense!
(∀ f )S (f ) = limn→∞S n(f ) and (∀ n) ∞−∞ ∣S n(f )∣df < ∞
⇒F−1
{S
(f
)} = limn→∞
F−1
{S n
(f
)}.
Digital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 9 / 106
8/21/2019 EC 2351 Digital Communication
10/106
Appendix: Extended Fourier transform
Since lima↓0 e −a
∣f
∣sgn(f ) = sgn(f ),lima↓0 ∞−∞ e −a∣f ∣sgn(f )e ı2πft df = lima↓0 −
0
−∞e f (a+ ı2πt )df +
∞
0
e f (−a+ ı2πt )df = lima↓0 − 1a + ı 2πt + 1a − ı 2πt
= lima↓0 ı 4πt
a2 + 4π2t 2 = ⎧⎪⎪⎨⎪⎪⎩0 t
= 0
ı 1πt t ≠ 0Hence, F −1 {2u −1(f )} = F −1 {1} +F −1 {sgn(f )} = δ (t ) + ı 1πt .
Digital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 10 / 106
F ( ) ( )
8/21/2019 EC 2351 Digital Communication
11/106
From x +(t ) to x (t )
X (f )
f 0f 0 − W f 0 + W −f 0−f 0 + W −f 0 − W ⇒
X (f )
0−W W
We then observeX (f ) = 2X +(f + f 0).
This implies
x (t ) = F −1{X (f )}= F −1{2X +(f + f 0)}
= 2x +
(t
)e − ı2πf 0 t
= (x
(t
) + ı x̂
(t
))e − ı2πf 0 t
Digital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 11 / 106
8/21/2019 EC 2351 Digital Communication
12/106
As a result,
x
(t
) + ı x̂
(t
) = x (
t
)e ı2πf 0t
which gives:
x
(t
) = Re
{x
(t
) + ı x̂
(t
)} = Re
x
(t
)e ı2πf 0t
By x (t ) = x i (t ) + ı x q (t ),
x (t ) = = Re (x i (t ) + ı x q (t ))e ı2πf 0t = ×i (t ) cos(2πf 0t ) − x q (t ) sin(2πf 0t )Digital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 12 / 106
F X (f ) X (f )
8/21/2019 EC 2351 Digital Communication
13/106
From X (f ) to X (f )From x
(t
) = Re
{x
(t
)e ı2πf 0 t
}, we obtain
X (f ) = ∞−∞ x (t )e − ı2πft dt
= ∞
−∞Re
x
(t
)e ı2πf 0t
e − ı2πft dt
= ∞−∞ 12 x (t )e ı2πf 0t + x (t )e ı2πf 0t ∗ e − ı2πft dt = 12
∞
−∞x (t )e − ı2π(f −f 0)t dt
+1
2 ∞
−∞x ∗ (t )e
− ı2π(f +f 0)t dt
= 12 [X (f − f 0) +X ∗ (−f − f 0)]
X ∗ (−f ) = ∫
∞−∞ (x (t )e
− ı2π(−f )t
)∗
dt = ∫ ∞−∞
x ∗ (f )e
− ı2πft dt Digital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 13 / 106
S
8/21/2019 EC 2351 Digital Communication
14/106
Summary
Terminologies & relations
Bandpass signal⎧⎪⎪⎨⎪⎪⎩
x
(t
) = Re
{x
(t
)e ı2πf 0t
}X
(f
) = 12
X
(f
−f 0
) +X ∗
(−f
−f 0
)Analytic signal or pre-envelope x +(t ) and X +(f )Lowpass equivalent signal or complex envelope
⎧⎪⎪⎨⎪⎪⎩x (t ) = (x (t ) + ı x̂ (t ))e − ı2πf 0 t X (f ) = 2X (f + f 0)u −1(f + f 0)Digital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 14 / 106
U f l t k
8/21/2019 EC 2351 Digital Communication
15/106
Useful to know
Terminologies & relations
From x (t ) = x i (t ) + ı x q (t ) = (x (t ) + ı x̂ (t ))e − ı2πf 0 t ,
⎧⎪⎪⎨⎪⎪⎩x i
(t
) = Re
(x
(t
) + ı x̂
(t
)) e − ı2πf 0t
x q (t ) = Im (x (t ) + ı x̂ (t )) e − ı2πf 0t Also from x (t ) = (x (t ) + ı x̂ (t ))e − ı2πf 0 t ,
⎧⎪⎪⎨⎪⎪⎩x
(t
) = Re
x (
t
) e ı2πf 0t
x̂ (t ) = Im x (t ) e ı2πf 0t Digital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 15 / 106
U f l t k
8/21/2019 EC 2351 Digital Communication
16/106
Useful to know
Terminologies & relations
From x (t ) = x i (t ) + ı x q (t ) = (x (t ) + ı x̂ (t ))e − ı2πf 0 t ,
⎧⎪⎪⎨⎪⎪⎩x i
(t
) = Re
(x
(t
) + ı x̂
(t
)) e − ı2πf 0t
x q (t ) = Im (x (t ) + ı x̂ (t )) e − ı2πf 0t Also from x (t ) = (x (t ) + ı x̂ (t ))e − ı2πf 0 t ,
⎧⎪⎪⎨⎪⎪⎩x
(t
) = Re
(x i (
t
) + ı x q (
t
)) e ı2πf 0t
x̂ (t ) = Im (x i (t ) + ı x q (t )) e ı2πf 0t Digital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 16 / 106
Us f l t k
8/21/2019 EC 2351 Digital Communication
17/106
Useful to know
Terminologies & relations
pre-envelope x +(t )complex envelope x (t )envelope
∣ ×
(t
)∣ =
x 2i
(t
) +x 2q
(t
) = r
(t
)
phase θ
(t
) = arctan
[x q
(t
)/x i
(t
)]Digital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 17 / 106 Modulaor/demodulator and Hilbert transformer
8/21/2019 EC 2351 Digital Communication
18/106
Modulaor/demodulator and Hilbert transformer
Usually, we will modulate and demodulator with respect to
carrier frequency f c , which may not be equal to the centerfrequency f 0.
x
(t
) → x
(t
) = Re
{x
(t
)e ı2πf c t
} = modulation
x (t ) → x (t ) = (x (t ) + ı x̂ (t ))e − ı2πf c t = demodulationThe modulation requires to generate x̂ (t ), a Hilberttransform of x (t )
Digital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 18 / 106
8/21/2019 EC 2351 Digital Communication
19/106
Hilbert transform is basically a 90-degree phase shifter.
H (f ) = F 1πt = − ı sgn(f ) = ⎧⎪⎪⎪⎨⎪⎪⎪⎩ −ı , f > 00, f = 0
ı , f 00 f 0−X (f ) f 0−1 f
8/21/2019 EC 2351 Digital Communication
20/106
Energy considerations
Definition (Energy of a signal)
The energy E s of a (complex) signal s (t ) is E s = ∞−∞ ∣s (t )∣2 dt Hence,
E x = ∞−∞ ∣x (t )∣2 dt
E x +
= ∞
−∞ ∣x +
(t
)∣2
dt
E x = ∞−∞ ∣x (t )∣2 dt We are interested in the connections among
Ex ,
Ex
+, and
Ex .
Digital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 20 / 106
8/21/2019 EC 2351 Digital Communication
21/106
From Parseval’s Theorem we see
E x = ∞−∞ ∣x (t )∣2 dt =
∞−∞ ∣X (f )∣2 df
Parsevals theorem: ∫ ∞−∞ x (t )y ∗(t )dt = ∫ ∞−∞ X (f )Y ∗(f )df Rayleigh’s theorem: ∫ ∞−∞ ∣x (t )∣2dt = ∫ ∞−∞ ∣X (f )∣2df Secondly
X (f ) = 12 X (f − f c )=X +(f )
+ 12X ∗ (−f − f c )=X ∗+ (−f )
Thirdly, f c ≫ W andX (f − f c )X ∗ (−f − f c ) = 4X +(f )X ∗+ (−f ) = 0 for all f Digital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 21 / 106
8/21/2019 EC 2351 Digital Communication
22/106
It then shows
E x
= ∞
−∞ 1
2X
(f
−f c
) + 1
2X ∗
(−f
−f c
)2
df
= 14E x + 14E x = 12E x
and
E x = ∞−∞ ∣X +(f ) +X ∗+ (−f )∣2 df = E x + + E x + = 2E x +Theorem (Energy considerations)E x = 2E x = 4E x +
Digital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 22 / 106
Extension of energy considerations
8/21/2019 EC 2351 Digital Communication
23/106
Extension of energy considerations
Definition (Inner product)We define the inner product of two (complex) signals x (t ) and
y (t ) as ⟨x
(t
), y
(t
)⟩ = ∞
−∞
x
(t
) y ∗
(t
)dt .
Parseval’s relation immediately gives
⟨x
(t
), y
(t
)⟩ = ⟨X
(f
), Y
(f
)⟩.
E x = ⟨x (t ), x (t )⟩ = ⟨X (f ), X (f )⟩E x = ⟨x (t ), x (t )⟩ = ⟨X (f ), X (f )⟩Digital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 23 / 106
8/21/2019 EC 2351 Digital Communication
24/106
We can similarly prove that
⟨x
(t
), y
(t
)⟩= ⟨X (f ), Y (f )⟩
= 1
2X
(f
−f c
) + 1
2X ∗
(−f
−f c
), 1
2Y
(f
−f c
) + 1
2Y ∗
(−f
−f c
)= 1
4 ⟨X (f − f c ), Y (f − f c )⟩ + 1
4 ⟨X (f − f c ), Y ∗ (−f − f c )⟩=0
+1
4
⟨X ∗
(−f
−f c
), Y
(f
−f c
)⟩=0 +1
4
⟨X ∗
(−f
−f c
), Y ∗
(−f
−f c
)⟩= 14⟨x (t ), y (t )⟩ + 1
4(⟨x (t ), y (t )⟩)∗ = 12 Re{⟨x (t ), y (t )⟩} .
Digital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 24 / 106
Corss-correlation of two signals
8/21/2019 EC 2351 Digital Communication
25/106
Corss correlation of two signals
Definition (Cross-correlation)
The cross-correlation of two signals x (t ) and y (t ) is defined as ρx ,y
= ⟨x (t ), y (t )⟩
⟨x
(t
), x
(t
)⟩ ⟨ y
(t
), y
(t
)⟩ = ⟨x (t ), y (t )⟩ E
x
E y
.
Definition (Orthogonality)
Two signals x
(t
) and y
(t
) are said to be orthogonal if ρx ,y
= 0.
The previous slide then shows ρx ,y = Re {ρx ,y }.ρx ,y 0
⇒ ρx ,y 0 but ρx ,y 0
/⇒ ρx ,y 0
Digital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 25 / 106
2 1-4 Lowpass equivalence of a bandpass system
8/21/2019 EC 2351 Digital Communication
26/106
2.1 4 Lowpass equivalence of a bandpass system
Definition (Bandpass system)
A bandpass system is an LTI system with real impulse response h(t ) whose transfer function is located around afrequency f c
Using a similar concept, we set the lowpass equivalentimpulse response as
h
(t
) = Re
h
(t
)e ı2πf c t
and H (f ) = 12 [H (f − f c ) +H ∗ (−f − f c )]
Digital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 26 / 106
Baseband input-output relation
8/21/2019 EC 2351 Digital Communication
27/106
Baseband input output relation
Let x
(t
) be a bandpass input signal and let
y (t ) = h(t )⋆x (t ) or equivalently Y (f ) = H (f )X (f )Then, we knowx
(t
) = Re
x
(t
)e ı2πf c t
h(t ) = Reh(t )e ı2πf c t
y (t ) = Re y (t )e ı2πf c t and
Theorem (Baseband input-output relation)
y (t ) = h(t ) ⋆ x (t ) ⇐⇒ y (t ) = 12
h(t ) ⋆ x (t )Digital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 27 / 106
8/21/2019 EC 2351 Digital Communication
28/106
Proof:
For f ≠ −f c (or specifically, for u −1(f + f c ) = u 2
−1(f + f c )),Note 12 = u −1(0) ≠ u 2−1(0) = 14 .Y
(f
) = 2Y
(f
+f c
)u −1
(f
+f c
)= 2H (f + f c )X (f + f c )u −1(f + f c )= 12 [2H (f + f c )u −1(f + f c ) ] ⋅ [2X (f + f c )u −1(f + f c )]
= 1
2H
(f
) ⋅X
(f
)The case for f = −f c is valid since Y (−f c ) = X (−f c ) = 0.
Digital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 28 / 106
8/21/2019 EC 2351 Digital Communication
29/106
The above applies to deterministic system. How aboutstochastic system?
x (t ) y (t ) h(t) ⇓
X(t ) Y(t ) h(
t
)
The text abuses the notation by using X (f ) as the spectrumof x
(t
) but using X
(t
) as the stochastic counterpart of x
(t
).
Digital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 29 / 106
8/21/2019 EC 2351 Digital Communication
30/106
2.7 Random processes
Digital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 30 / 106
Random Process
8/21/2019 EC 2351 Digital Communication
31/106
DefinitionA random process is a set of indexed random variables
{X
(t
), t
∈ T }, where
T is often called the index set.
Classification
1 If T is a finite set ⇒ Random Vector2 If
T =Z or Z+
⇒ Discrete Random Process
3 If T = R or R+ ⇒ Continuous Random Process4 If T = R2,Z2,⋯,Rn,Zn ⇒ Random FieldDigital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 31 / 106
Examples of random process
8/21/2019 EC 2351 Digital Communication
32/106
p p
Example
Let U be a random variable uniformly distributed over [−π, π).ThenX(t ) = cos (2πf c t +U)
is a (continuous) random process.
Example
Let B be a random variable taking values in
{−1, 1
}. Then
X(t ) = cos(2πf c t ) if B = −1sin(2πf c t ) if B = +1 = cos 2πf c t − π4 (B + 1)is a (continuous) random process.
Digital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 32 / 106
Statistical properties of random process
8/21/2019 EC 2351 Digital Communication
33/106
p p p
For any integer k
> 0 and any t 1, t 2,
⋯, t k
∈ T , the
finite-dimensional cumulative distribution function (cdf) for
X(t ) is given by:F X (t 1,⋯, t k ; x 1,⋯, x k ) = Pr {X(t 1) ≤ x 1,⋯, X(t k ) ≤ x k }
As event
[X
(t
) ≤ ∞] (resp.
[X
(t
) ≤ −∞]) is always regarded
as true (resp. false),
limx s →∞F X
(t 1,
⋯, t k ; x 1,
⋯, x k
)= F X (t 1,⋯, t s −1, t s +1, t k ; x 1,⋯, x s −1, x s +1,⋯, x k )andlim
x s →−∞F X
(t 1, , t k ; x 1, , x k
) 0
Digital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 33 / 106
8/21/2019 EC 2351 Digital Communication
34/106
Definition
Let X(t ) be a random process; then the mean function is mX(t ) = E[X(t )],
the (auto)correlation function is
R X(t 1, t 2) = E [X(t 1)X∗(t 2)] ,and the (auto)covariance function is
K X(t 1, t 2) = E (X(t 1) −mX(t 1)) (X(t 2) −mX(t 2))∗ Digital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 34 / 106
Definition
8/21/2019 EC 2351 Digital Communication
35/106
Let X(t ) and Y(t ) be two random processes; then the cross-correlation function is
R X,Y(t 1, t 2) = E [X(t 1)Y∗(t 2)] ,and cross-covariance function is
K X,Y(t 1, t 2) = E (X(t 1) −mX(t 1)] [Y(t 2) −mY(t 2))∗ Proposition
R X,Y(t 1, t 2) = K X,Y(t 1, t 2) +mX(t 1)m∗Y(t 2)R Y,X(t 2, t 1) = R ∗X,Y(t 1, t 2) R X(t 2, t 1) = R ∗X(t 1, t 2)K Y,X
(t 2, t 1
) K ∗X,Y
(t 1, t 2
) K X
(t 2, t 1
) K ∗X
(t 1, t 2
)Digital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 35 / 106 Stationary random processes
8/21/2019 EC 2351 Digital Communication
36/106
Definition
A random process X(t ) is said to be strictly or strict-sense stationary (SSS) if its finite-dimensional joint distributionfunction is shift-invariant, i.e. for any integer k > 0, any t 1,
⋯, t k
∈ T and any τ ,
F X (t 1 −
τ,
⋯, t k −
τ ; x 1,⋯, x k ) =
F X (t 1,⋯
, t k ; x 1,⋯, x k )Definition
A random process X
(t
) is said to be weakly or wide-sense
stationary (WSS) if its mean function and (auto)correlation
function are shift-invariant, i.e. for any t 1, t 2 ∈ T and any τ ,mX(t − τ ) = mX(t ) and R X(t 1 − τ, t 2 − τ ) = R X(t 1, t 2).The above condition is equivalent to
mX
(t
) constant and R X
(t 1, t 2
) R X
(t 1 t 2
).
Digital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 36 / 106
Wide-sense stationary random processes
8/21/2019 EC 2351 Digital Communication
37/106
Definition
Two random processes X(t ) and Y(t ) are said to be jointly wide-sense stationary if
Both X
(t
) and Y
(t
) are WSS;
R X,Y(t 1, t 2) = R X,Y(t 1 − t 2).Proposition
For jointly WSS X
(t
) and Y
(t
),
R Y,X(t 2, t 1) = R ∗X,Y(t 1, t 2) ⇒ R X,Y(τ ) = R ∗Y,X(−τ )K Y,X(t 2, t 1) = K ∗X,Y(t 1, t 2) ⇒ K X,Y(τ ) = K ∗Y,X(−τ )
Digital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 37 / 106
Gaussian random process
8/21/2019 EC 2351 Digital Communication
38/106
Definition
A random process {X(t ), t ∈ T } is said to be Gaussian if for any integer k > 0 and for any t 1,⋯, t k ∈ T , the finite-dimensional joint cdf
F X(t 1,⋯, t k ; x 1,⋯, x k ) = Pr [X(t 1) ≤ x 1,⋯, X(t k ) ≤ x k ]is Gaussian.
RemarkThe joint cdf of a Gaussian process is fully determined by itsmean function and its (auto)covariance function.
Digital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 38 / 106
Gaussian random process
8/21/2019 EC 2351 Digital Communication
39/106
Definition
Two real random processes {X(t ), t ∈ T X } and {Y(t ), t ∈ T Y }are said to be jointly Gaussian if for any integers j , k > 0 and for any s 1,
⋯, s j
∈ T X and t 1,
⋯, t k
∈ T Y, the finite-dimensional
joint cdf
Pr [X(s 1) ≤ x 1,⋯, X(s j ) ≤ x j , Y(t 1) ≤ y 1,⋯, Y(t k ) ≤ y k ]is Gaussian.
DefinitionA complex process is Gaussian if the real and imaginary processes are jointly Gaussian.
Digital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 39 / 106
Gaussian random process
8/21/2019 EC 2351 Digital Communication
40/106
Remark
For joint (in general complex) Gaussian processes,“uncorrelatedness”, defined as
R X,Y
(t 1, t 2
) = E
[X
(t 1
)Y∗
(t 2
)]= E[X(t 1)]E[Y∗(t 2)] = mX(t 1)m∗Y(t 2),implies “independence”, i.e.,
Pr [X(s 1) ≤ x 1,⋯, X(s j ) ≤ x j , Y(t 1) ≤ y 1,⋯, Y(t k ) ≤ y k ]= Pr [X(s 1) ≤ x 1,⋯, X(s k ) ≤ x k ]⋅Pr [Y(t 1) ≤ y 1,⋯, Y(t k ) ≤ y k ]Digital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 40 / 106
Theorem
8/21/2019 EC 2351 Digital Communication
41/106
If a Gaussian random process X(t ) is WSS, then it is SSS.Proof:
For any k > 0, consider the sampled random vector
⃗Xk
=⎡⎢⎢⎢⎢⎢⎢⎢⎣
X(t 1)X
(t 2
)⋮X(t k )⎤⎥⎥⎥⎥⎥⎥⎥⎦
The mean vector and covariance matrix of
⃗Xk are respectively
m⃗Xk = E[⃗Xk ] = ⎡⎢⎢⎢⎢⎢E[X(t 1)]E[X(t 2)]⋮E
[X
(t k
)]
⎤⎥⎥⎥⎥⎥ = mX(0) ⋅ ⃗1Digital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 41 / 106
8/21/2019 EC 2351 Digital Communication
42/106
and
K ⃗X = E[⃗Xk ⃗XH k ] = ⎡⎢⎢⎢⎢⎢⎣K X
(0
) K X
(t 1
−t 2
) ⋯K X(t 2 − t 1) K X(0) ⋯⋮ ⋮ ⋱⎤⎥⎥⎥⎥⎥⎦It can be shown that for a new sampled random vector
⎡⎢⎢⎢⎢⎢⎢⎢⎣
X(t 1 + τ )X(t 2 + τ )
⋮X
(t k
+τ
)
⎤⎥⎥⎥⎥⎥⎥⎥⎦the mean vector and covariance matrix remain the same.
Hence, X(t ) is SSS.Digital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 42 / 106
Power spectral density
8/21/2019 EC 2351 Digital Communication
43/106
Definition
Let R X(τ ) be the correlation function of a WSS randomprocess X(t ). The power spectral density (PSD) or power spectrum of X
(t
) is defined as
S X(f ) = ∞−∞ R X(τ )e − ı2πf τ d τ
Let R X,Y(τ ) be the cross-correlation function of two jointly WSS random process X
(t
) and Y
(t
); then the cross spectral
density (CSD) is
S X,Y(f ) = ∞−∞ R X,Y(τ )e − ı2πf τ d τ Digital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 43 / 106
Properties of PSD
8/21/2019 EC 2351 Digital Communication
44/106
PSD (in units of watts per Hz) describes thedistribution/density of power as a function of frequency.
Analogously, probability density function (pdf) describesthe distribution/density of probability as a function of
outcome.The integration of PSD gives power of the randomprocess over the considered range of frequency.Analogously, the integration of pdf gives probability over
the considered range of outcome.
Digital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 44 / 106
Theorem
8/21/2019 EC 2351 Digital Communication
45/106
Theorem
S X
(f
) is non-negative and real (which matches that the power
of a signal cannot be negative).
Proof: S X(f ) is real becauseS X
(f
) = ∞
−∞
R X
(τ
)e − ı2πf τ d τ
= ∞−∞ R X(−s )e ı2πfs ds (s = −τ )
= ∞
−∞R ∗X
(s
)e ı2πfs ds
= ∞−∞ R X(s )e − ı2πfs ds ∗= S ∗X(f )Digital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 45 / 106
S X(f ) is non-negative because of the following (we only provethis based on that T ⊂ R and X(t) = 0 outside [−T , T ]).
8/21/2019 EC 2351 Digital Communication
46/106
this based on that T R and X(t ) 0 outside [ T , T ]).S X
(f
) = ∞−∞
E
[X
(t
+τ
)X∗
(t
)]e − ı2πf τ d τ
= EX∗(t ) ∞−∞ X(t + τ )e − ı2πf τ d τ (s = t + τ )
= E
X∗
(t
) ∞−∞
X
(s
)e − ı2πf (s −t ) ds
= E X∗(t )X̃(f )e ı2πft In notation, X̃(f ) = F{X(t )}Since the above is a constant with respect to t (by WSS),S X
(f
) = 1
2T
T
−T E
X∗
(t
)X̃
(f
)e ı2πft
dt
= 12T E X̃(f ) T −T X∗(t )e ı2πft dt 1
2T E
X̃
(f
)X̃∗
(f
) 1
2T E
∣X̃
(f
)∣2
0.
Digital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 46 / 106
Wiener-Khintchine theorem
8/21/2019 EC 2351 Digital Communication
47/106
Theorem (Wiener-Khintchine)
Let {X(t ), t ∈ R} be a WSS random process. Define XT (t ) = X(t ) if t ∈ [−T , T ]0, otherwise.
and set
X̃T (f ) = ∞−∞ XT (t )e − ı2πft dt = T −T X(t )e − ı2πft dt .If S X(f ) exists (i.e., R X(τ ) has a Fourier transform), then
S X(f ) = limT →∞
1
2T EX̃T (f )2
Digital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 47 / 106
Variations of PSD definitions
8/21/2019 EC 2351 Digital Communication
48/106
Power density spectrum : Alternative definitionFourier transform of auto-covariance function (e.g.,Robert M. Gray and Lee D. Davisson, RandomProcesses: A Mathematical Approach for Engineers,p. 193)
I remark that from the viewpoint of digitalcommunications, the text’s definition is more appropriatesince
the auto-covariance function is independent of a
mean-shift; however, random signals with differentmeans consume different “powers.”
Digital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 48 / 106
8/21/2019 EC 2351 Digital Communication
49/106
What can we say about, e.g., the PSD of stochasticsystem input and output?
x (t )x
(t
) y (t )
y
(t
) h(t )
12 h
(t
)
⎧⎪⎪⎨⎪⎪⎩
◻(t ) = Re{◻(t )e ı2πf c t }
◻
(t
) = (◻(t
) + ı ˆ
◻(t
))e − ı2πf c t
where “
◻” can be x , y or h.
⇓X
(t
)X(t ) Y
(t
)Y(t ) h
(t
)1
2
h(t )
⎧⎪⎪⎨⎪⎪⎩◻(
t
) = Re
{◻
(t
)e ı2πf c t
}◻
(t
) = (◻(t
) + ı ˆ
◻(t
))e − ı2πf c t
where “◻” can be X, Y or h.
Digital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 49 / 106
8/21/2019 EC 2351 Digital Communication
50/106
2.9 Bandpass and lowpass random processes
Digital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 50 / 106
Definition (Bandpass random signal)
A b d (WSS) t h ti i l X(t) i l d
8/21/2019 EC 2351 Digital Communication
51/106
A bandpass (WSS) stochastic signal X(t ) is a real randomprocess whose PSD is located around central frequency f 0, i.e.
S X(f ) = 0 for all ∣f ± f 0∣ > W
S X
(f
)f 0f 0
−W f 0
+W
−f 0
−f 0
+W
−f 0
−W
f 0 may not be thecarrier frequency f c !
We know
⎧⎪⎪⎨
X(t ) = Re {X(t )e ı2πft }X t X t ı X̂ t e − ı2πf 0 t
Digital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 51 / 106
Assumption
8/21/2019 EC 2351 Digital Communication
52/106
The bandpass signal X
(t
) is WSS.
In addition, its complex lowpass equivalent process X
(t
) is
WSS. In other words,Xi (t ) and Xq (t ) are WSS.Xi
(t
) and Xq
(t
) are jointly WSS.
Under this fundamental assumption, we obtain thefollowing properties:
P1) If X
(t
) zero-mean, both Xi
(t
) and Xq
(t
) zero-mean
because mX = mXi cos(2πf c t ) −mXq sin(2πf c t ) .P2)
⎧⎪⎪⎨
R Xi (τ ) = R Xq (τ )R Xi ,Xq τ R Xq ,Xi τ
Digital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 52 / 106
Proof of P2):
R
8/21/2019 EC 2351 Digital Communication
53/106
R X
(τ
)= E
[X
(t
+τ
)X
(t
)]= E ReX(t + τ )e ı2πf c (t +τ )ReX(t )e ı2πf c t = E [(Xi (t + τ ) cos(2πf c (t + τ )) −Xq (t + τ ) sin(2πf c (t + τ )))(
Xi
(t
)cos
(2πf c t
) −Xq
(t
)sin
(2πf c t
))]= R Xi (
τ
) +R Xq (
τ
)2 cos(2πf c τ )+R Xi ,Xq (τ ) −R Xq ,Xi (τ )2
sin(2πf c τ )+
R Xi
(τ
) −R Xq
(τ
)2 cos(2πf c (2t + τ )) (= 0)−R Xi ,Xq (τ ) +R Xq ,Xi (τ )2
sin(2πf c (2t + τ )) (= 0)Digital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 53 / 106
P3) R X(τ ) = Re 12 R X(τ )e ı2πf c τ .
8/21/2019 EC 2351 Digital Communication
54/106
) X( ) 2 X( ) Proof : Observe from P2),
R X(τ ) = E [X(t + τ )X∗ (t )]= E [(Xi (t + τ ) + ı Xq (t + τ ))(Xi (t ) − ı Xq (t ))]=
R Xi
(τ
) +R Xq
(τ
) − ı R Xi ,Xq
(τ
) + ı R Xq ,Xi
(τ
)= 2R Xi (τ ) + ı 2R Xq ,Xi (τ ).Hence, also from P2),R X
(τ
) = R Xi
(τ
)cos
(2πf c t
) −R Xq ,Xi
(τ
)sin
(2πf c t
)= Re12 R X(τ )e ı2πf c τ Digital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 54 / 106
8/21/2019 EC 2351 Digital Communication
55/106
P4) S X
(f
) = 14
S X
(f
−f c
) +S ∗X
(−f
−f c
).
Proof : A direct consequence of P3).
◻Note :Amplitude X̃(f ) = 12 X̃(f − f c ) + X̃∗ (−f − f c )Amplitude sequare
∣X̃(f )∣2 = 14 X̃(f − f c ) + X̃∗ (−f − f c )2
= 14 ∣X̃(f − f c )∣2 + X̃∗ (−f − f c )
2
Wiener-Khintchine: S X(f ) ≡ ∣X̃(f )∣2.
Digital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 55 / 106
8/21/2019 EC 2351 Digital Communication
56/106
P5) Xi
(t
) and Xq
(t
) uncorrelated if one of them has
zero-mean.Proof : From P2),
R Xi ,Xq
(τ
) = −R Xq ,Xi
(τ
) = −R Xi ,Xq
(−τ
).
Hence, R Xi ,Xq (0) = 0 (i.e.,E[Xi (t )Xq (t )] = 0 = E[Xi (t )]E[Xq (t )]).
◻Digital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 56 / 106
8/21/2019 EC 2351 Digital Communication
57/106
P6) If S X
(−f
) = S ∗X
(f
), Xi
(t
+τ
) and Xq
(t
) uncorrelated for
any τ if one of them has zero-mean.
Proof : From P3),
R X
(τ
) = 2R Xi
(τ
) − ı 2R Xq ,Xi
(τ
).
S X(−f ) = S ∗X(f ) implies R X(τ ) is real;hence, R Xq ,Xi (τ ) = 0 for any τ . ◻
We next discuss the PSD of a system.
Digital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 57 / 106
X(t ) Y(t ) h(t )
Y(t ) = ∞−∞ h(τ )X(t − τ )d τ ∞
h d
8/21/2019 EC 2351 Digital Communication
58/106
mY
= mX
−∞
h
(τ
)d τ
R X,Y(τ ) = EX(t + τ ) ∞−∞ h(u )X(t − u )du ∗
= ∞
−∞h∗
(u
)R X
(τ
+u
)du
= ∞
−∞h∗
(−v
)R X
(τ
−v
)dv
= R X(τ ) ⋆ h∗(−τ )R Y(τ ) = E
∞
−∞h(u )X(t + τ − u )du
∞
−∞h(v )X(t − v )dv ∗
= ∞
−∞ h(u ) ∞
−∞ h∗(v )R X((τ − u ) + v )dv du = ∞−∞ h(u )R X,Y(τ − u )du R X,Y τ h τ R X τ h
∗ τ h τ .
Digital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 58 / 106
8/21/2019 EC 2351 Digital Communication
59/106
Thus,
S X,Y(f ) = S X(f )H ∗(f ) since ∞−∞ h∗(−τ )e − ı2πf τ d τ = H ∗(f )and
S Y(f ) = S X,Y(f )H (f ) = S X(f )∣H (f )∣2.
Digital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 59 / 106
White process
8/21/2019 EC 2351 Digital Communication
60/106
Definition (White process)
A (WSS) process W(t ) is called a white process if its PSD is constant for all frequencies:S W
(f
) = N 0
2
This constant is usually denoted by N 02 because the PSDis two-sided. So, the power spectral density is actually N 0
per Hz (N 0/2 at f = −f 0 and N 0/2 at f = f 0).The autocorrelation function R W(τ ) = N 02 δ (⋅), where δ (⋅)is the Dirac delta function.
Digital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 60 / 106
Why negative frequency?
A “sample answer”:
8/21/2019 EC 2351 Digital Communication
61/106
A sample answer :
It is just an imaginary convenient way created by Human
to correspond to the “imaginary” domain of a complexsignal (that is why we call it “imaginary part”).
By giving respectively the spectrum for f 0 and
−f 0 (which
may not be symmetric), we can tell the amount of real
part and imaginary part in time domain corresponding tothis frequency.
For example, if the spectrum is conjugate symmetric, weknow imaginary part (in time domain)
= 0.
Notably, in communications, imaginary part is the partthat will be modulated by (or transmitted with carrier)sin(2πf c t ); on the contrary, real part is the part that willbe modulated by (or transmitted with carrier) cos 2πf c t .
Digital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 61 / 106
Why δ (⋅) function?D fi i i (Di d l f i )
8/21/2019 EC 2351 Digital Communication
62/106
Definition (Dirac delta function)
Define the Dirac delta function δ
(t
) as
δ (t ) = ∞, t = 0;0, t
≠ 0
,
which satisfies the replication property , i.e., for every continuous point of g (t ),g
(t
) = ∞−∞
g
(τ
)δ
(t
−τ
)d τ.
Hence, by replication property,
∞−∞
δ u du
∞−∞
δ t τ d τ
∞−∞
1 δ t τ d τ 1.
Digital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 62 / 106
Not that it seems δ (t ) = 2δ (t ) = ∞, t = 0;0, t ≠ 0 ; but with
8/21/2019 EC 2351 Digital Communication
63/106
g 1
(t
) = 1 and g 2
(t
) = 2 continuous at all points,
1 = ∞−∞ g 1(t )δ (t )dt ≠ ∞−∞ g 2(t )δ (t )dt = 2.So, mathematician does not “like” this function as itcontradicts intuition:
f (t ) = g (t ) for t ∈ R except for countably many points
⇒ ∞
−∞
f
(t
)dt
= ∞
−∞
g
(t
)dt
if
∞
−∞
f
(t
)dt is finite
.
Hence, δ (t ) and 2δ (t ) are two “different” Diract deltafunctions by definitions. (The multiplicative constantcannot be omitted!) Very “artificial” indeed.
Digital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 63 / 106
Comment: x a y a x y is incorrect if a
8/21/2019 EC 2351 Digital Communication
64/106
Comment: x
+a
= y
+a
⇒ x
= y is incorrect if a
= ∞.
As a result, saying
∞ = ∞ (or δ
(t
) = 2δ
(t
) ) is not a
“rigorously defined” statement.
Summary: The Dirac delta function, like “
∞”, is simply
a concept defined only through its replication property .
Hence, a white process W(t ) that has autocorrelationfunction R W
(τ
) = N 0
2 δ
(τ
) is just a convenient and
simplified notion for theoretical research about real world
phenomenon. Usually, N 0 = KT , where T is the ambienttemperature in kelvins and k is Boltzman’s constant.Digital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 64 / 106
Discrete-time random processes
The property of a time-discrete process X n n Z+
8/21/2019 EC 2351 Digital Communication
65/106
The property of a time discrete process
{X
[n
], n
∈Z
}can be “obtained” using sampling notion via the Dirac
delta function.X[n] = X(nT ), a sample at t = nT from atime-continuous process X(t ), where we assume T = 1for convenience.The autocorrelation function of a time-discrete process isgiven by:
R X[m] = E{X[n +m]X[n]}= E
{X
(n
+m
)X
(n
)}= R X(m), a sample from R X(t ).
R X(0)R X(1)
R X(2)R X(3)R X(4)R X(5)
R X(6)R X(7)R X(8)R X(9)
R X(10)R X(11)
R X(12)R X(13)
Digital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 65 / 106
8/21/2019 EC 2351 Digital Communication
66/106
S X
[f
] = ∞
−∞ ∞
n=−∞R X
(t
)δ
(t
−n
)e − ı2πft dt
= ∞n=−∞
∞−∞ R X(t )e − ı2πft δ (t − n)dt
= ∞
n=−∞R X
(n
)e − ı2πfn (Replication Property)
= ∞n=−∞
R X[n]e − ı2πfn (Fourier Series)Hence, by Fourier sesies,
R X[n] = 12−12 S X[f ]e ı2πfmdf = R X(n) = ∞−∞ S X(f )e ı2πfmdf .Digital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 66 / 106
8/21/2019 EC 2351 Digital Communication
67/106
2.8 Series expansion of random processes
Digital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 67 / 106
2.8-1 Sampling bandLimited random process
8/21/2019 EC 2351 Digital Communication
68/106
Deterministic case
A deterministic signal x (t ) is called band-limited if X (f ) = 0 for all ∣f ∣ > W Shannon-Nyquist theorem: x
(t
) can be perfectly
reconstructed if the sampling rate f s > 2W andx (t ) = ∞
n=−∞x n
2W sinc 2W t − n
2W
Note that the above is only sufficient, not necessary.
Digital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 68 / 106
Stochastic case
A WSS h i X i id b b d li i d
8/21/2019 EC 2351 Digital Communication
69/106
A WSS stochastic process X
(t
) is said to be band-limited
if its PSD S X(f
) = 0 for all
∣f
∣ > W
It follows that
R X
(τ
) = ∞
n=−∞R X
n
2W sinc
2W
τ
− n
2W In fact, the random process X(t ) can be reconstructed byits (random) samples in the sense of mean square.
Theorem
E X(t ) − ∞n=−∞
X n2W sinc 2W t − n
2W 2 = 0
Digital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 69 / 106
The random samples
Problems of using these random samples
8/21/2019 EC 2351 Digital Communication
70/106
Problems of using these random samples.
These random samples X n2W ∞n=−∞
are in general
correlated unless X(t ) is zero-mean white.EX n
2W X∗ m
2W = R X n − m
2W
≠ E
X
n
2W E
X∗
m
2W = mX
2.
If X(t ) is zero-mean white,EX n
2W X∗ m
2W = R X n − m
2W = N 0
2 δ n − m
2W
= EX n2W EX∗ m2W = mX2 = 0 except n = m.
Thus, we will introduce the uncorrelated KL expansionsin slide 87.
Digital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 70 / 106
8/21/2019 EC 2351 Digital Communication
71/106
2.9 Bandpass and lowpass random processes (revisited)
Digital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 71 / 106
Definition (Filtered white noise)
A N i ll d fil d hi i if i PSD l
8/21/2019 EC 2351 Digital Communication
72/106
A process N
(t
) is called a filtered white noise if its PSD equals
S N(f ) = N 02 , ∣f ± f c ∣
8/21/2019 EC 2351 Digital Communication
73/106
P0 1) By fundamental assumption on slide 52, we obtainthat X
(t
) and X̂
(t
) are jointly WSS.
R X,X̂(τ ) and R ̂X(τ ) are only functions of τ because X̂(t ) is theHilbert transform of X(t ), i.e., R
X,X̂(τ ) = R X(τ ) ⋆ h∗(−τ ) =
−R X(τ )⋆h(τ ) (since h∗(−τ ) = −h(τ )) and R ̂X(τ ) = R X,X̂(τ )⋆h(τ ).P0-2) X
i (t ) = Re (X(t ) + ı X̂(t ))e −
ı2πf c t
is WSS byfundamental assumption.P2 ′
) ⎧⎪⎪⎨⎪⎪⎩
R X
(τ
) = R ̂X
(τ
)R X,X̂(
τ
) = −R ̂X,X(
τ
)(X(t ) + ı X̂(t ) is the “lowpassequivalent” signal of Xi (t )!)(Xi (
t
)+ı Xq (
t
) is the lowpass
equivalent signal of X(t )!)Also, R ̂X,X(τ ) = R̂ X(τ ), where R̂ X(τ ) is the Hilbert transformoutput due to input R X τ .
Digital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 73 / 106
P3 ′) R Xi (τ ) = Re 12 R (X+ ı X̂)(τ )e − ı2πf c τ 1 2 f
8/21/2019 EC 2351 Digital Communication
74/106
R Xi
(τ
) = Re
1
2R (X+ ı X̂)
(τ
)e − ı2πf c τ
= Re(R X(τ ) + ı R ̂X,X(τ ))e − ı2πf c τ = R X(τ ) cos(2πf c τ ) + R̂ X(τ ) sin(2πf c τ )Note that Ŝ X
(f
) = S X
(f
)H Hilbert
(f
) = S X
(f
)(−ı sgn
(f
)).
P4 ′) S Xi (f ) = S X(f − f c ) + S X(f + f c ) = S Xq (f )) for ∣f ∣
8/21/2019 EC 2351 Digital Communication
75/106
Terminologies & relations
● ⎧⎪⎪⎪⎪⎪⎨⎪⎪⎪⎪⎪⎩
R X(τ ) = Re 12 R X(τ ) e ı2πf c τ (P 3)R ̂X,X(τ ) = R X(τ ) ⋆ hHilbert(τ )
P0-1
= Im 12 R X(τ ) e ı2πf c τ ● Then: 12 R X(τ ) = R Xi (τ ) + ı R Xq ,Xi (τ )
Proof of P 3
= (R X(τ ) + ı R ̂X,X(τ ))e − ı2πf c τ
● ⎧⎪⎪⎪⎪⎨⎪⎪⎪⎪R Xi (τ ) = Re (R X(τ ) + ı R ̂X,X(τ )) e − ı2πf c τ (P 3′)
R Xq ,Xi (τ ) = Im (R X(τ ) + ı R ̂X,X(τ )) e − ı2πf c τ = R Xi (τ ) ⋆ hHilbert(τ )Digital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 75 / 106
8/21/2019 EC 2351 Digital Communication
76/106
Proof: Hence,
R Xq ,Xi (τ ) = Im(R X(τ ) + ı R ̂X,X(τ ))e − ı2πf c τ = −
R X
(τ
)sin
(2πf c τ
) +R ̂X,X
(τ
)cos
(2πf c τ
)= −R X(τ ) sin(2πf c τ ) + ˆR X(τ ) cos(2πf c τ ).The property can be proved similarly to P4 ′).
Digital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 76 / 106
8/21/2019 EC 2351 Digital Communication
77/106
2.2 Signal space representation
Digital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 77 / 106
Key idea & motivation
8/21/2019 EC 2351 Digital Communication
78/106
The low-pass equivalent representation removes thedependence of system performance analysis on carrierfrequency.
Equivalent vectorization of the (discrete or continuous)signals further removes the “waveform” redundancy inthe analysis of system performance.
Digital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 78 / 106
Vector space concepts
I d t n ∗ H
8/21/2019 EC 2351 Digital Communication
79/106
Inner product:
⟨v1, v2
⟩ = ∑ni =1 v 1,i v ∗2,i
= vH 2 v1
(“H ” denotes Hermitian transpose)
Orthogonal if ⟨v1, v2⟩ = 0Norm:
∥v
∥ = ⟨v, v
⟩Orthonormal: ⟨v1, v2⟩ = 0 and ∥v1∥ = ∥v2∥ = 1Linearly independent:
k i =1
ai vi = 0 iff ai = 0 for all i Digital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 79 / 106
Vector space concepts
Triangle inequality
8/21/2019 EC 2351 Digital Communication
80/106
∥v1 + v2∥ ≤ ∥v1∥ + ∥v2∥Cauchy-Schwartz inequality∣⟨
v1, v2
⟩∣ ≤ ∥v1
∥ ∥v2
∥.
Equality holds iff v1 = av2 for some aNorm square of sum:
∥v1
+v2
∥2
= ∥v1
∥2
+ ∥v2
∥2
+ ⟨v1, v2
⟩ + ⟨v2, v1
⟩Pythagorean: if ⟨v1, v2⟩ = 0, thenv1 v2
2v1
2v2
2
Digital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 80 / 106
Eigen-decomposition
8/21/2019 EC 2351 Digital Communication
81/106
1 Matrix transformation w.r.t. matrix A
v̂ = Av2 Eigenvalues of square matrix A are solutions
{λ
} of
characteristic polynomialdet(A − λI ) = 0
3 Eigenvectors for eigenvalue λ is solution v of
Av = λvDigital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 81 / 106
Signal space concept
How to extend the signal space concept to a (complex)
8/21/2019 EC 2351 Digital Communication
82/106
How to extend the signal space concept to a (complex)function/signal z
(t
) defined over
[0, T
) ?
Answer: We can start by defining the inner product forcomplex functions.
Inner product: ⟨z 1(t ), z 2(t )⟩ = ∫ T 0 z 1(t )z ∗2 (t )dt Orthogonal if ⟨z 1(t ), z 2(t )⟩ = 0.Norm:
∥z
(t
)∥ = ⟨z
(t
), z
(t
)⟩Orthonormal: ⟨z 1(t ), z 2(t )⟩ = 0 and ∥z 1(t )∥ = ∥z 2(t )∥ = 1.Linearly independent: ∑k i =1 ai z i (t ) = 0 iff ai = 0 for allai ∈ C
Digital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 82 / 106
Triangle Inequality
z 1 t z 2 t z 1 t z 2 t
8/21/2019 EC 2351 Digital Communication
83/106
∥ ( ) + ( )∥ ≤ ∥ ( )∥ + ∥ ( )∥Cauchy Schwartz inequality∣⟨z 1(t ), z 2(t )⟩∣ ≤ ∥z 1(t )∥ ⋅ ∥z 2(t )∥Equality holds iff z 1
(t
) = a
⋅z 2
(t
) for some a
∈C
Norm square of sum:
∥z 1(t ) + z 2(t )∥2 = ∥z 1(t )∥2 + ∥z 2(t )∥2+ ⟨
z 1
(t
), z 2
(t
) ⟩+ ⟨z 2
(t
), z 1
(t
)⟩Pythagorean property: if ⟨z 1(t ), z 2(t )⟩ = 0,z 1 t z 2 t
2z 1 t
2z 2 t
2
Digital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 83 / 106
8/21/2019 EC 2351 Digital Communication
84/106
Transformation w.r.t. a function C
(t , s
)ẑ (t ) = T
0C (t , s )z (s )ds
This is in parallel to
v̂ v̂ t = ns =1
At ,s v s = Av.
Digital Communications: Chapter 2 Ver 2012.09.29 Po-Ning Chen 84 / 106
Eigenvalues and eigenfunctions
8/21/2019 EC 2351 Digital Communication
85/106
Given a complex continuous function C (t , s ) over [0, T )2, theeigenvalues and eigenfunctions are
{λk
} and
{ϕk
(t
)} such
that
T 0
C (t , s )ϕk (s )ds = λk ϕk (t ) (In parallel to Av = λv)
Digital Communications: Chapter 2 Ver 2012.09.29 Po-Ning Chen 85 / 106
Mercer’s theorem
8/21/2019 EC 2351 Digital Communication
86/106
Theorem (Mercer’s theorem)Give a complex continuous function C (t , s ) over [0, T ]2 that is Hermitian symmetric (i.e., C
(t , s
) = C ∗
(s , t
)) and
nonnegative definite (i.e.,
∑i
∑ j ai C
(t i , t j
)a∗ j
≥ 0 for any
{ai
}and {t i }). Then the eigenvalues {λk } are reals, and C (t , s )has the following eigen-decompositionC
(t , s
) = ∞
k =1λk ϕk
(t
)ϕ∗k
(s
).
Digital Communications: Chapter 2 Ver 2012.09.29 Po-Ning Chen 86 / 106
Karhunen-Loève theorem
Theorem (Karhunen-Loève theorem)
8/21/2019 EC 2351 Digital Communication
87/106
Let
{Z
(t
), t
∈ [0, T
)} be a zero-mean random process with a
continuous autocorrelation function R Z(t , s ) = E[Z(t )Z∗(s )].Then Z(t ) can be written as Z
(t
) M2
= ∑∞k =1 Zk
⋅ϕk
(t
) 0
≤ t
< T
where “ =” is in the sense of mean-square,Zk = ⟨Z(t ), ϕk (t )⟩ = ∫ T 0 Z(t )ϕ∗k (t )dt and {ϕk (t )} are orthonormal eigenfunctions of R Z(t , s ).
Merit of KL expansion: {Zk } are uncorrelated. (Butsamples {Z(k /(2W ))} are not uncorrelated even if Z(t )is bandlimited!)
Digital Communications: Chapter 2 Ver 2012 09 29 Po-Ning Chen 87 / 106
Proof.
8/21/2019 EC 2351 Digital Communication
88/106
E[Zi Z∗ j ] = E T 0 Z(t )ϕ∗i (t )dt T 0 Z(s )ϕ∗ j (s )ds ∗
= T
0
T
0R Z
(t , s
)ϕ j
(s
)ds
ϕ∗i
(t
)dt
= T 0 λ j ϕ j (t )ϕ∗i (t )dt
= ⎧⎪⎪⎨⎪⎪⎩
λ j if i = j 0
=E
[Zi
]E
[Z∗ j
] if i
≠ j
Digital Communications: Chapter 2 Ver 2012 09 29 Po-Ning Chen 88 / 106
LemmaFro a given orthonormal set {φk (t )}, how to minimize the energy of error signal e t s t ŝ t for ŝ t spanned by (i d li bi ti f) φ t ?
8/21/2019 EC 2351 Digital Communication
89/106
( ) = ( ) − ( ) ( )(i.e., expressed as a linear combination of)
{φk
(t
)}?
Assume ŝ (t ) = ∑k ak φk (t ); then
∥e
(t
)∥2
= ∥s
(t
) −ŝ
(t
)∥2
= ∥s
(t
) −∑k ∶k ≠i ak φk
(t
) −ai φi
(t
)∥2
= ∥s (t ) −∑k ∶k ≠i ak φk (t )∥2 + ∥ai φi (t )∥2−⟨s (t ) −∑k ∶k ≠i ak φk (t ), ai φi (t )⟩− ⟨
ai φi
(t
), s
(t
) −∑k ∶k ≠i ak φk
(t
)⟩= ∥s (t ) −∑k ∶k ≠i ak φk (t )∥2
+ ∣ai ∣2
−a∗i ⟨s (t ), φi (t )⟩ − ai ⟨φi (t ), s (t )⟩By taking derivative w.r.t. Re ai and Im ai , we obtainai s t , φi t .
Digital Communications: Chapter 2 Ver 2012 09 29 Po-Ning Chen 89 / 106
Definition
If every finite energy signal s t (i.e., s t 2
) satisfies
8/21/2019 EC 2351 Digital Communication
90/106
y gy g
( )(
∥ ( )∥ < ∞)
∥e (t )∥2 = s (t ) −k
⟨s (t ), φk (t )⟩φk (t )2 = 0(equivalently,
s (t ) L2= k
⟨s (t ), φk (t )⟩φk (t ) = k
ak ⋅ φk (t )(in the sense that the norm of the difference between
left-hand-side and right-hand-side is zero), then the set of orthonormal functions {φk (t )} is said to be complete .
Digital Communications: Chapter 2 Ver 2012 09 29 Po Ning Chen 90 / 106
Example (Fourier series)⎧⎪⎪ 2T
cos2πkt
T ,
2
T sin
2πkt
T 0 k Z
⎫⎪⎪
8/21/2019 EC 2351 Digital Communication
91/106
⎨⎪⎪⎩
∶ ≤ ∈ ⎬⎪⎪⎭is a complete orthonormal set for signals defined over [0, T )with finite number of discontinuities.
For a complete orthonormal basis, the energy of s
(t
) is
equal to ∥s (t )∥2 = j
a j φ j (t ),k
ak φk (t )= j k
a j a∗k
⟨φ j
(t
), φk
(t
)⟩= j
a j a∗
j
j
a j 2
Digital Communications: Chapter 2 Ver 2012 09 29 Po Ning Chen 91 / 106
Given a deterministic function s (t ), and a set of completeorthonormal basis φk t (possibly countably infinite),
8/21/2019 EC 2351 Digital Communication
92/106
{ ( )}s
(t
) can be written as
s (t ) L2= ∞k =0
ak φk (t ), 0 ≤ t ≤ T where
ak = ⟨s (t ), φk (t )⟩ = T 0
s (t )φ∗k (t )dt .In addition,
∥s (t )∥2 = k ∣ak ∣2.Digital Communications: Chapter 2 Ver 2012 09 29 Po Ning Chen 92 / 106
Remark
8/21/2019 EC 2351 Digital Communication
93/106
Remark
In terms of energy (and error rate):A bandpass signal s (t ) can be equivalently “analyzed”through lowpass equivalent signal s
(t
) without the
burden of carrier freq f c ;
A lowpass equivalent signal s (t ) can be equivalently“analyzed through (countably many) {ak = ⟨s (t ), φk (t )⟩}without the burden of continuous waveforms.
Digital Communications: Chapter 2 Ver 2012.09
.29 Po Ning Chen 93 / 106
Gram-Schmidt procedure
Given a set of functions v 1 t , v 2 t , , v k t
8/21/2019 EC 2351 Digital Communication
94/106
( ) ( ) ⋯ ( )1 φ1(t ) = v 1(t )∥v 1(t )∥2 Compute for i
= 2, 3,
⋯, k (or until
∥φi
(t
)∥ = 0),
γ i (t ) = v i (t ) − i
−1
j =1 ⟨v i (t ), φ j (t )⟩φ j (t )and set φi
(t
) = γ i (t )∥γ i (t )∥ .
This gives an orthonormal basis φ1(t ), φ2(t ),⋯, φk ′(t ), wherek ′ ≤ k .
Digital Communications: Chapter 2 Ver 2012.09
.29 Po Ning Chen 94 / 106
Example
Find a Gram-Schmidt orthonormal basis of the following signals
8/21/2019 EC 2351 Digital Communication
95/106
signals.
Digital Communications: Chapter 2 Ver 2012.09
.29 Po Ning Chen 95 / 106
Sol.φ1(t ) = v 1(t )∥v 1(t )∥ = v 1(t )√ 2
8/21/2019 EC 2351 Digital Communication
96/106
γ 2(t ) = v 2(t ) − ⟨v 2(t ), φ1(t )⟩φ1(t )= v 2(t ) − 30
v 2(t )φ∗1(t )dt φ1(t ) = v 2(t )Hence φ
2(t ) = γ 2(t )∥γ 2(t )∥ =
v 2(t )√ 2
.
γ 3
(t
) = v 3
(t
) − ⟨v 3
(t
), φ1
(t
)⟩φ1
(t
) − ⟨v 3
(t
), φ2
(t
)⟩φ2
(t
)= v 3(t ) −√ 2φ1(t ) − 0 ⋅ φ2(t ) = −1, 2 ≤ t
8/21/2019 EC 2351 Digital Communication
97/106
γ 4(t ) = v 4(t ) − ⟨v 4(t ), φ1(t )⟩φ1(t ) − ⟨v 4(t ), φ2(t )⟩φ2(t )−⟨v 4(t ), φ3(t )⟩φ3(t )=
v 4
(t
)−(−√ 2
)φ1
(t
) − (0
)φ2
(t
) −φ3
(t
) = 0
Orthonormal basis=
{φ1
(t
), φ2
(t
), φ3
(t
)}, where 3
≤ 4.
Digital Co icatio s Cha te 2 Ve 2012.09
.29 Po Ni g Che 97 / 106
Example Represent the signals in slide 95 in terms of the orthonormal basis obtained in the same example.
8/21/2019 EC 2351 Digital Communication
98/106
Sol.
v 1
(t
) = √
2φ1
(t
) +0
⋅φ2
(t
) +0
⋅φ3
(t
) ⇒ √
2, 0, 0
v 2(t ) = 0 ⋅ φ1(t ) +√ 2 ⋅ φ2(t ) + 0 ⋅ φ3(t ) ⇒ 0,√ 2, 0v 3(t ) = √ 2φ1(t ) + 0 ⋅ φ2(t ) + ⋅φ3(t ) ⇒ √ 2, 0, 1v 4
(t
) = −√
2φ1
(t
) +0
⋅φ2
(t
) +1
⋅φ3
(t
) ⇒ −√
2, 0, 1
The vectors are named signal space representations orconstellations of the signals.
Di it l C i ti Ch t 2 V 2012.
09.
29 P Ni Ch 98 / 106
Remark
The orthonormal basis is not unique.For example for k 1 2 3 re define
8/21/2019 EC 2351 Digital Communication
99/106
For example, for k
= 1, 2, 3, re-define
φk (t ) = 1, k − 1 ≤ t
8/21/2019 EC 2351 Digital Communication
100/106
s 1(t ) ⇒ (a1, a2,⋯, an) for some complete basiss 2(t ) ⇒ (b 1, b 2,⋯, b n) for the same complete basisd 12 = Euclidean distance between s 1(t ) and s 2(t )= n
i =1(ai − b i )2
= ∥s 1(t ) − s 2(t )∥ = T
0 ∣s 1(t ) − s 2(t )∣2dt Di it l C i ti Ch t 2 V 2012
.
09.
29 P Ni Ch 100 / 106
Bandpass and lowpass orthonormal basis
Now let’s change our focus from 0, T to ,
8/21/2019 EC 2351 Digital Communication
101/106
[ ) (−∞ ∞)A time-limited signal cannot be bandlimited to W .A bandlimited signal cannot be time-limited to T .
Hence, in order to talk about the ideal bandlimited signal, wehave to deal with signals with unlimited time span.
Re-define the inner product as:
⟨f (t ), g (t )⟩ = ∞−∞ f (t )g ∗(t )dt Digital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 101 / 106
Let s 1,(t ) and s 2,(t ) be lowpass equivalent signals of thebandpass s 1 t and s 2 t
8/21/2019 EC 2351 Digital Communication
102/106
( ) ( )S 1,(f ) = S 2,(f ) = 0 for ∣f ∣ > f B s i (t ) = Re s i ,(t )e ı2πf c t where f c ≫ f B Then, as we have proved in slide 24,
⟨s 1(t ), s 2(t )⟩ = 12 Re {⟨s 1,(t ), s 2,(t )⟩} .Proposition
If ⟨s 1,(t ), s 2,(t )⟩ = 0, then ⟨s 1(t ), s 2(t )⟩ = 0.Digital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 102 / 106
PropositionIf {φn,(t )} is a complete basis for the set of lowpass signals,then
φn t Re 2φn, t e ı2πf c t
8/21/2019 EC 2351 Digital Communication
103/106
⎧⎪⎪⎪⎪⎨⎪⎪⎪⎪⎩ ( ) = √ ( ) φ̃n(t ) = −Im√ 2φn,(t ) e ı2πf c t = Re ı√ 2φn,(t ) e ı2πf c t is a complete orthonormal set for the set of bandpass signals.
Proof: First, orthonormality can be proved by
⟨φn
(t
), φm
(t
)⟩ = 1
2Re
√
2φn,
(t
),√
2φm,
(t
) =
⎧⎪⎪⎨⎪⎪⎩
1 n = m0 n
≠ m
φ̃n(t ), φ̃m(t ) = 12
Re ı√ 2φn,(t ), ı√ 2φm,(t ) = ⎧⎪⎪⎨1 n = m0 n m
Digital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 103 / 106
and φ̃n(t ), φm(t ) = 12
Re ı√ 2φn,(t ),√ 2φm,(t ) Re ı φn, t , φm, t
8/21/2019 EC 2351 Digital Communication
104/106
= { ⟨ ( ) ( )⟩}= ⎧⎪⎪⎨⎪⎪⎩Re{ ı} = 0 n = m0 n ≠ m
Now, with⎧⎪⎪⎪⎪⎪⎪⎪⎪⎨⎪⎪⎪⎪⎪⎪⎪⎪⎩
s (t ) = Re {s (t )e ı2πf c t }ŝ (t ) = Re {ŝ (t )e ı2πf c t }ŝ
(t
) L2
= n
an,φn,
(t
) with an,
= ⟨s
(t
), φn,
(t
)⟩∥s (t ) − ŝ (t )∥2
= 0we haves t ŝ t 2
1
2s t ŝ t
2 0
Digital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 104 / 106
and
ŝ t Re ŝ t e ı2πf c t
8/21/2019 EC 2351 Digital Communication
105/106
( ) = ( ) = Ren an,φn,(t )e ı2πf c t
= n
Re
an,
√ 2
Re
√ 2φn,
(t
)e ı2πf c t
+Iman,√ 2 Im−√ 2φn,(t ) e ı2πf c t
= n Re
an,
√ 2φn
(t
) +Im
an,
√ 2 φ̃n
(t
) ◻Digital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 105 / 106
What you learn from Chapter 2
8/21/2019 EC 2351 Digital Communication
106/106
Random processWSSautocorrelation and crosscorrelation functionsPSD and CSD
White and filtered white
Relation between (bandlimited) bandpass and lowpassequivalent deterministic signalsRelation between (bandlimited) bandpass and lowpass
equivalent random signalsProperties of autocorrelation and power spectrum density
Role of Hilbert transform
Signal space conceptDigital Communications: Chapter 2 Ver. 2012.09.29 Po-Ning Chen 106 / 106