Chapter 4 :
Puf :
5
[xHY]
[=
E <XIHX*H] · E [x( + z)x + (t - )]
-
Rx(d)xRx(0)
Questions : 1) Mean
-
-
2) Autocorvitation
Example 7
-
:
XH =
Acs)anfot 4) +
)
&
A constant
f[0 hit]
,
p(q)
11255////I 2T
EH] = EL _
Acos)anfot d)] +
= AE
[fot+] +
fla
=
(f()) Pale)ay
A
=(anfot y)dyIndependent + = 0= mx
of t
* Rx)t,+ - z) = E[XH(* (t 81] -
= El AcolanfotA coslanfo (t-61)
[c/bitfo(at-6)us/an88))
as(a+b) 1000
-
k)
las a cash =
=
2
>
A cf0G) T) Independent
of t => WSS
-
Ge
-f .
(at-2) +
24
Rx (8) = clatfor
SxIf) = <61y-fol 61f + +
fol]
-A4
-fo fo
Example2(2nfot) -
BSinsaif-t)
And B centered independent random variables with the
2
are same variances+
↑
↑ T TB +2
ElAB) EIA) E(B)
= =
Mean = 0 = .
CIAT-ElBl 0 =
[x)t]
= =
C. Cfo-) Sinarfo = Endopendlet
of time
Rx(t , + 2) -
= E(x()X* (t -
z1) =
z() (s)myot) -
Bsin(aif
-
+)] (Aus/any
-
. +2)
BSin/Lifolt-sl)
GAI-EIB... --
= EIA) coslanfot) cos(hig St-8) . + ElB
Sindaifot)(sinaitflt-2)-
EA) = + Im = =
r
ElB = = Pr = +
3
= Rx(t , +8) =
/anfolt-81) ott
inf + Sindanfot) Sindaitfolt-81) +
=
[cosifor] independent
of
+ > WSS
&p : as l
ash + Sin
& a sink = cosa -
b)
Rx (2) = +
as(aifot)
Example 3 :
cyclostationarity
Acslaif t) E(XIH] E[XSt-T1] 1
XH = .
=
,
i=
2fs
ACentered EIAI = 0 Rx(f ,+ -
z) =
Rx(
+ -
T
,+ T
- -
z)
EIA ? = T+ Ima = +" +O
HEXHI] =
ECAcoskifotI] = EIAl .
cofot If twasntcentertime)
=>
cos(aifotXX =0 (In this wes
* Rx(t , + b) -
=
E@X(H)y
* (t -
21) =
ElAcoslargot Acostanfo H-81)]
s
-ELA") (slaigot Chif It--all
To
:
.
Examples
4 :
X(H =
A cos/aif+ b) +
A centered
,
PS ,
indofd
Can-814
9) :
O
sit
y) Pa14104
·
EX(H] =
EXAcoslaifot + $1]m(cossa fo + +
+] = hig +
Eltfot
= =0 d
Independent
of time
even
if EA] was to
* Rx (t ECAcos(aifot d) Acclanfo (t - d1]
,
+ -) = + +
=
ECA'] EceSarifot $) Cs(hitf + + 2) H1] rep + losb .
-
+ : los a
↓ [Cos(a b)
+ + ((a -
b)]
=
ELA] [Glangv) 1 (21 % (n + 2) 201] + o
-
+
]
"
-2) + 2)
E(at
ELA
= cos
katfoo =0
claif) ind
of WS
Slide 15 :
Exh we wantNo minimize the error
: >
we predicted that X( + to)
+ = < X 14
Y(t + to) =
XoX(t) we went x1f+tol thun adreh shi fot
J
XH
E(xt + to) - X (t + tol)
1) MSE = E [(x (t to)
*
+ -
x(x) + 139]
XX]
*
-
E[Xto) Xt + "X-240
CNote :
=
EXX Rx(0) + 20" Rx1d -
200 Rx(tol )
=
Rx(te-ta) = Was
MS =
0 = 20Rd-2R
Rx(o)
20 =
Rx(d
M Strin R(0)-RIto *
Rx (tol Rx(0) 1
If No MSO
= =
;
=
Rx(d) * | t to) X(+ + to)
+ =
X(H =
to > -
lim RxSto) =
Imx)" ,
In this example the Mean =8
to >+D
-
If Rx(to) =0
= > MSEmin = Rx(0)
Slide 142
1)
2) MSE = ESIX-aY- bl]
=
E[x] + Ely h"-2aEIXY]-2bE[X] + +Lab ERy]
MSE min :
E
GaEFy"] -cE[XY] 2 bE[y] =0
[ CMS
+
2b -
2E[X] + 2aE[y] = 0
cust
Y
=> a
= b EbX]
YE2]
-
;
=
[(X-mxl]
<
with Is" = E = E[x] EEX] -
E[x] E[X]EET]
and Y /correlation weff)
-
match
To My
perfect limes
& MSEmin = (1 -
(2)
#if =
11 (41) MStmin = 0
HEf = 0 MSEmin =T (a = 0
,
b =
mx)
Slids 20 : power spectral density Ss(f)
NolL
Splf) =
N If -f
· Autocorrelation
function :
Rb(z) =
No G(t) ↑
Slid 27 ECYCH] = E[XiC] = E(XIHX(H] = Rx(o)
independent of t
(tal] By Hitz) WSS
*
et[X(t)
↓ X =
in slids at
(t + 8) use the formula
By ,
-
2)] E[X()x() X(t z)X(t 2)
-
E[yH) yt (t
-
- =
4C .
GRV
= E[XIH XIt1] · E [xlt-2(x(t-81) + 2E [XH X It- b)]
Et X(H)X( + 81] -
Pyt Ri(0) + 2 R? (2)
of
=
inde t = WSS
Power spectral
density
Sylf) >
Ry(z)
=
Ry (0) <(f) + 2Sx (f) Sx1f)
Sylf) =
+
imple
Sxlf) )
:
-
B -
B12 B12 + B
Slide
-i
28
E [yH] E [A
my(f) H(f 0) hIt]
= = . mx
+ =
[Shlul X(t-uldu]
J DE [hu) Xt-uoT
+
=> =
/hu) u)] du
it's because
WSS mean doesn't
=
mxhudu may
% =
Initial value theorem
=
with time
change
Yalt-2) ]
*
Slide 2 :
Ry , Y g (81 E[yn (H)
=
-E[Sh1(u)X1(t-ul de She (2) X 202]
*
(t-8) -
]du
Shlu)he E[X(u)
*
=
(2)
t2
f)h + (ulha(n) Pxx2(8 +v
-
v) dude
Let w = - 2
() hr(u)h(n- w)Rx1x2(8 -
w) dudw
Let g(w) = hn(w) + h * (- w) =
(hr(y)h(n w) -
du
=>
Rysye 181 =
Sglw) Rxix (8-w) du
=> (2) g(8) + Rx x2(8)
Ryn ya =
,
h (b) + h( 8) Ry
= +
xz(z) -
Slide 30 :
Ryx(2) = h(z) * 1A Rx(z)
= h(z) * Rx( %)
[
H
Rx(b)
+
Rxy(f) = 1xh ( 2) -
+
Application
- yet)
i X(H)
- -
h(z)
Ryx(z) = +1* Rx(8)
= h(z) +
Nf(z) =
Nh(z)
hi =
v BzEYS
Elim yx*-o