Beruflich Dokumente
Kultur Dokumente
Spectrum Communications
JungLang Yu
Department of Electronic Engineering
Fu Jen Catholic University
Taipei, Taiwan
Tel: +866-2-29052102
Fax:+866-2-29042638
e-mail : yujl@ee.fju.edu.tw
Ftp:
http://www.ee.fju.edu.tw/communication/main.html
OUTLINE
TEXT BOOKS
Textbook:
1. R.L. Peterson, et. al., Introduction to spread spectrum
communications, 1995,
2. V.P. Ipatov, Spread spectrum and CDMA, Principles and
Applications,, 2005
Reference:
1. IS-95 CDMA and CDMA 2000 Cellular/PCS Systems Implementation,
V.K. Grag, 2000, Chap 1~Chap7
2. S.G. Glisic, Adaptive WCDMA, 2003.
3. S. Verdu, Multi-user Detection, 1998
4. P. Van Rooyen, et., al., Space-time processing for CDMA mobile
communications, 2000
5. A.J. Viterbi,Principles of spread spectrum communication, 1995
Introductions
Basic digital communication concepts
Spread spectrum systems
Binary shift-register sequences
Synchronization, code acquisition, code tracking
Performance of spread spectrum systems in jamming
environments
Code division multiple access channel
Single user matched filter
Optimum multi-user detection, Decorrelating Detector,
Nondecorrelating linear multi-user detection
Multi-Carrier CDMA (OFDM)
GRADE
Computer Exercise 35%
Midterm project and presentation 30% (paper survey) : 23th
April
Term project and presentation 35% (duplicate its simulation
results) : 11th June
Due date
z Computer exercise: a week after assignment
z Midterm project : on 2nd April
z Term project : on 21th May
Project topics:
Topics :
z Multi-user detection
z RAKE receivers
z Channel Estimation
z Synchronization in spread spectrum systems
z Power control in CDMA systems
z Handover techniques
z Wireless LANs based on spread spectrum technology
z Advanced wireless techniques: OFDM, UWB, etc.
z Any topics related CDMA are OK after discussing with Dr. Yu.
References :
z IEEE Transaction on Communication
z IEEE Transaction on Vehicle Technology
z IEEE Transaction on Signal Processing
z IEEE Transaction on Wireless Communication
z Signal Processing
TD-SCDMA
It is the combination of TDMA system and synchronization
CDMA, which is proposed by the Simens (Germany) and
Datang (China) in 1999.
3G Telecommunication licenses
z
z
z
0.7 Reference
1.
2.
3.
4.
5.
6.
7.
8.
9.
10.
11.
12.
13.
Reference
1.
2.
1.1.
1.2.
1.2.1.
y (t ) = s1 (t ) + n(t ) or s2 (t ) + n(t )
n(t ) N (0, N 0 / 2)
1/20
2/20
Error probability :
Send
s1 (t )
T
Send
s2 (t )
T
where
T
E12 = s1 (t ) s2 (t )dt
0
Ei = si2 (t )dt ,
i = 1, 2
E[W] = 0,
T
N0
2
( s2 (t ) s1 (t )) 2 dt =
N0
( E1 + E2 2 E12 )
2
3/20
and
Pe = Q (
d /2
) = Q(
E1 + E2 2 E12
E + E2 2 E12
) = Q( 1
)
2 n
2N0
Define
z=
( E1 + E2 ) Eb
=
2N0
N0
R12 =
2 E12
E
= 12
E1 + E2 Eb
Then
Pe = Q ( z (1 R12 ))
Example:
Bandwidth:
For a rectangular pulse of period T, the spectrum is
S ( f ) = A2Tsinc( fT )
Sm ( f ) =
A2T
{sinc[( f f 0 )T ] + sinc[( f + f 0 )T ]}
2
4/20
BWRF =
2
T
2.5
T
Bandwidth efficiency:
R
1
R=
,
BW
T
for baseband signal
1
for ASK,PSK
= 1/ 2
1/ 2.5 for FSK
1.2.2.
z
z
z
Bandpass filter : correlator receiver or matched filter with cos(.) and sin(.).
Envelope detector : square law detector.
1
E
Pe e z / 2 for z = b >> 1
2
N0
FSK:
if 1 is sent
A cos( wct + ) + n (t )
r (t ) =
A cos( wc t + wt + ) + n(t ) if 0 is sent
1
Pe = e z / 2 for all z
2
z
DPSK:
Transmitter
5/20
a)
A2 T
A2T
cos(
)
dt
=
cos( i )
i
i 1
2 0
2
Pe = Q ( z )
=
b)
Optimum Rx
6/20
r (t ) = A cos( wc t + i + ) + n(t )
AT
cos( i + )
0
2
T
AT
yi = A cos( wc t + i + )sin( wc t )dt =
sin( i + )
0
2
T
y = xi xi 1 + yi yi 1 =
Pe =
1.2.3.
A2T 2
A2T 2
cos( i i 1 ) =
cos( i )
4
4
1 z
e
2
Performance Comparison
7/20
8/20
{i (t ), i = 1" K ,0 t T }, K M
by
v1 (t )
v1
a)
v1 (t ) = s1 (t ), 1 (t ) =
b)
v2 (t ) = s2 (t ) < s2 ,1 > 1 (t ), 2 (t ) =
c)
d)
v2 ( t )
v2
v3 ( t )
v3
Where
T
x =
sn (t )
can be represented by
9/20
i =1
i =1
sn (t ) = (t )S n
1.3.2
If
s j (t )
is transmitted, then
r (t ) = s j (t ) + n(t ) = (t )S j + n(t )
T
Z i = Sij + N i
T
Statistics of Z i , N i
a)
N i Gaussian, so is Z i
b)
E [ N i ] = 0, E[ Z i ] = Sij
c)
N0
2
i* (t ) j (t )dt =
N0
ij
2
10/20
d)
cov( Z i , Z j ) = E[( Z i Z i )( Z j Z j )* ] = E [ N i N *j ] =
e)
i =1
i =1
1
2
K
2
f ( Z / s j (t )) = f ( Z i / s j (t )) = (2 ) e
= ( N 0 )
1.3.3
N0
2
K
2
( Z i Sij ) 2
i =1
N0
exp(
) = ( N 0 )
N0
ij
2
( Z i Sij )2
2 2
exp(
2
1
ZSj )
N0
M-ary PSK
si (t ) =
2 Es
cos( wc t + i ), 0 t Ts , i = 1" M
Ts
Where i =
2 (i 1)
, Es is symbol energy
M
2
2
si (t ) = Es cos( i )
cos( wc t ) sin( i )
sin( wc t )
Ts
Ts
= Es [ cos( i )1 (t ) sin( i )2 (t )]
If only two neighboring symbols are considered, e.g. s1 (t ), sM (t ) , then the BER is
computed by
P2 ( sM / s1 ) = Q (
Es sin(
11/20
)
M ) = Q ( 2 Es sin( ))
N0
M
2 Es
2 Es
sin( )) Ps 2Q (
sin( ))
N0
M
N0
M
Q(
1.3.4
M-ary FSK
si (t ) =
2 Es
cos( wc t + (i 1) wt ), 0 t Ts , i = 1" M
Ts
i ( t ) =
1
. We define basis as
Ts
2
cos( wc t + (i 1) wt ), 0 t Ts , i = 1" M
Ts
Then we have
si (t ) = Es i (t )
The SER has the upper bound of
M
i =2
i =2
Ps = P ( s2 " sM s1 ) P2 ( si s1 ) = Q (
2 Es 2
) = ( M 1)Q ( Es N 0 )
1.3.5
QAM
si (t ) =
2
( Ai cos wot + Bi sin wot ) = Ai1 (t ) + Bi2 (t ), 0 t Ts
Ts
Es
Es =
2( M 1) 2
a
3
1 2a 2
2a 2
2a 2
1
Q(
) + Q(
) + 2Q (
)( L 2) = 2(1 )Q ( 2a 2 N 0 )
L
L
12/20
13/20
1.4.1
R
BW
M-PSK,M-QAM,M-DPSK
BW =
R=
1
log 2 M (bps )
Ts
log 2 M
R
=
= 0.5log 2 M
BW
2
Coherent M-FSK
BW =
2
Ts
1 M 1 1 M + 3
+
+ =
Ts
2Ts
Ts
2Ts
R=
1
log 2 M (bps )
Ts
2log 2 M
R
=
BW
M +3
Noncoherent M-FSK
BW = M
2
Ts
R=
1
log 2 M (bps )
Ts
log 2 M
R
=
BW
2M
1.4.2
z
Es N 0 vs. Eb N 0
2 Ts
Ts = Tb log 2 M
Es = Eb log 2 M
E s Eb
log 2 M
=
N0 N0
Ps vs. Pb
Pb
Ps
log 2 M
for M-FSK
Pb
MPs
2( M 1)
14/20
15/20
z
z
z
1.5.1
N
k =1
Ak cos( wc t + m (t ) + k )
Scattering effect
sr (t ) = k =1 Ak cos( wc t + m (t ) + k )
N
16/20
sr (t )
can be expressed as
sr (t ) = R cos( wc t + m (t ) + )
Where
R= X +X
2
I
2
Q
Rayleigh distribution, f R ( r ) =
r2
2 2
= R1 cos( wc t + m (t ) + 1 )
Where
R1 Rician distribution, f R1 ( r1 ) =
r1
exp(
r12 + A2
)I0 (
Ar1
), r1 0
Rayleigh channel
1.5.2
Consider the coherent receiver (the frequency and phase of the received signal are known)
T
yD = [ R cos( wc t + m (t ) + ) + n(t )]
0
=R
2
cos( wc t + )dt
T
T
cos(m (T )) + W
2
T
N
cos(m (T )) indicates the signal term and W N (0, 0 ) is the noise term.
2
2
Further, the amplitude of cos(m (T )) for BPSK, FSK, and DPSK is unity. Then we can find
where R
17/20
z=
Eb R 2T
=
N0 2 N0
f ( z) =
1.5.3
R 2T
1
z
exp( ), z 0, z = E
z
z
2 N0
Pe ( z ) = Q ( 2 z )
However, z is random variable of chi-square distribution. We compute the averaged BER as
follows.
Pe = Pe ( z ) = Q( 2 z ) f ( z )dz =
2z
1
t2
1
z
exp( )dt exp( )dz
2
z
z
2
Define
u=
2z
du =
1
t2
1
z
exp( )dt , dv = exp( )dz
2
z
z
2
1
z
e z dz , v = exp( )
z
2 2 z
Therefore
z
z
1
Pe = Q( 2 z ) exp( ) ( exp( ))(
e z )dz
0
z 0
z
4 z
1
1
2 2
1
1
exp{ z (1 + )}dz
z
z
Let
w = z , dw =
dz
dz = 2wdw
2 z
Pe
18/20
becomes
Pe =
=
1
1
2 2
1 1
w exp{ w (1 + z )}2wdw
2
exp{ w (1 + z )}dw
2
Let
1
x2
1
x = 2 w2 (1 + ), or
= w2 (1 + )
z
2
z
Then
Pe =
=
1 1
exp{
0
1
1
2
(1 + 1 z )
x2
1
}(
)dx
2
2(1 + 1 z )
1
2
exp{
0
1
x2
z
)
}dx = (1
2
1+ z
2
1.5.4
Coherent FSK
The BER is given by
Pe ( z ) = Q( z )
Pe =
1
z
)
(1
2
2+ z
DPSK
The conditional BER is given by
Pe ( z ) =
1
exp( z )
2
19/20
1
1
1
z
Pe = Pe ( z ) = exp( z ) f ( z )dz = exp( z ) exp( )dz
2
2
z
z
0
0
1
1
1
1
1
=
exp{ z (1 + )}dz =
((1 + )1 ) exp{ z (1 + )}
2z 0
z
2z
z
z 0
=
1
2( z + 1)
Noncoherent FSK
The conditional BER is given by
1
z
Pe ( z ) = exp( )
2
2
Then the averaged BER is computed by
1
z
1
z 1
z
Pe = Pe ( z ) = exp( ) f ( z )dz = exp( ) exp( )dz
2
2
2
2 z
z
0
0
1
1 1
1
1 1
1 1
exp{ z ( + )}dz =
(( + ) 1 ) exp{ z ( + )}
2z 0
2 z
2z
2 z
2 z 0
1
z +2
20/20
1/16
Reference
1.
2.
V.P. Ipatov, Spread spectrum and CDMA, Principles and Applications, 2005, Chapter 2.4, 3.1,
3.2, 7.1
Introduction
2.1
z The modulation and demodulation of spread spectrum technique must have the following
characteristics
a) The transmitted signal energy must occupy a bandwidth which is much larger than the
information bit rate and which is approximately independent of the information bit rate.
b) Demodulation must be done in part by correlation of the received signal with a replica of the
signal used in the transmitter to spread the information signal.
Lowpass equivalent signals
2.2
z
(t ) is the phase. Bandpass signal can be expressed in terms of lowpass equivalent signal
by s (t ) = Re[ sl (t )e
jwct
Es =
1 2
1
1
2
a (t ) dt = sl (t ) dt = El
2
2
2
Therefore, we have
El = 2 Es
2.3
2.3.1
Pulse-noise jamming
2/27/2007
2/16
z A pulse-noise jammer having total averaged power J transmits pulses of bandlimited white
Gaussian noise whose PSD just covers the SS system bandwidth W (PSD= N J = J ). The
W
duty factor for the jammer is denoted by , which is the fraction of time during which the
NJ
jammer is ON. When the jammer is ON, the received jammer PSD is N J' = J
W =
T NJ
E = WN J T = WN J T N J = N J =
T
PE = Q( 2
Eb
)
N 0 + N J'
PE = Q ( 2
Eb
)
N0
Eb
Eb
) + Q ( 2
)
N0
N 0 + N J'
NJ
PE ( ) Q( 2
Eb
E
) = Q( 2 b )
'
NJ
NJ
2/27/2007
PE ( )
3/16
exp( x
x 2
)
2 , we have
exp( b )
NJ
4Eb N J
2 Eb
1
1
2e 2 Eb N J
z The following plot shows that the BERs of BPSK in AWGN (continuous-noise jammer) and
worse-case pulse-noise jammer.
2.3.2
z The goal of the LPD system is to use a signaling scheme that results in the minimum probability
of being detected within some time interval.
z The spread spectrum techniques can achieve the requirement of LPD.
2/27/2007
2.4
2.4.1
BPSK
4/16
z Transmitter
Sd(t)
d(t)
St(t)
2/27/2007
5/16
b p (t iT ), b {1}, p (t ) = ( T )
d (t ) =
i =
sd (t ) = 2 Pd (t ) cos( wo t ) = 2 P cos( wo t + d (t ))
c(t ) =
c p(t iT )
i
: spreading code, p (t ) = (
i =
t
), ci {1},
Tc
Sv ( f ) =
1
2
Sa ( f ) G( f ) ,
T
where
S a ( f ) = Ra (m) exp(2fmT )
m =
2/27/2007
Su ( f ) =
6/16
1
[S v ( f f o ) + S v ( f + f o )]
4
b) Baseband
d (t ) =
i =
Sb ( f ) = 1; P( f ) = Tsinc( fT )
PSD{d (t )} =
1
2
Sb ( f ) P( f ) = Tsinc 2 ( fT )
T
D( f )
c) Passband
sd (t ) = 2 Pd (t ) cos( wo t )
2P
[ D( f f o ) + D( f + f o )]
4
1
= PT sinc 2 ( f f o )T + sinc 2 ( f + f o )T
2
Sd ( f ) =
snew (t ) = 2 Pd (t )c(t )
S new ( f ) = 2 PTcsinc 2 ( fTc )
St ( f ) =
1
PTc sinc 2 ( f f o )Tc + sinc 2 ( f + f o )Tc
2
Sd(f)
2/27/2007
7/16
St(f)
y(t)
r(t)
)
y (t ) = r (t )c(t Td )
)
= 2 Pd (t Td ) cos( wo t + ) + 2 J c(t Td ) cos( wo t + J )
14444244443 14444244443
Low data rate BPSK modulation
1
PTc sinc 2 ( f f o )Tc + sinc 2 ( f + f o )Tc
2
1
+ J [ ( f f o ) + ( f + f o )]
2
1
S y ( f ) = PT sinc 2 ( f f o )T + sinc 2 ( f + f o )T
2
1
+ JTc sinc 2 ( f f o )Tc + sinc 2 ( f + f o )Tc
2
Sr ( f ) =
2/27/2007
8/16
Sr(f)
Sy(f)
f o +1 2T
J o = S J ( f ) H ( f ) df = 2 S J ( f ) df
f o 1 2T
1
= 2 JTc sinc 2 ( f f o )Tc df JTc 1df
2
JT
= c
T
z The signal power at the ideal BPF output has no changed. But the jammer power is reduced
JTc
. We define the processing gain of spread spectrum as
from J to
T
T
Gp =
Tc
2.4.2
Balanced QPSK
z Transmitter
s (t ) = P c1 (t ) cos( wo t + d (t )) P c2 (t ) sin( wo t + d (t ))
2/27/2007
d (t ) {
9/16
3
}
4
z Receiver
s (t Td ) = Pc1 (t Td )cos( wo t + d (t Td ) + )
Pc2 (t Td )sin( wo t + d (t Td ) + )
x(t ) =
1 s (t T )c (t T )2cos( w t + w t + )
d
1
d
o
IF
2
2.4.3
Dual-channel QPSK
2/27/2007
2.5
2.5.1
Coherent FHSS
10/16
z Transmitter
2/27/2007
b p (t iT ),
d (t ) =
p1 (t ) =
i =
11/16
(T )
s
sd (t ) = 2 Pd (t ) cos( wo t ) = 2 P cos( wo t + d (t ))
hT (t ) =
n =
is synthesized frequency
Tc > Ts
Tc < Ts
z Receiver
y1(t)
sr(t)
z The image reject filter is used to reject the image signal at frequency wn wo .
sr (t ) = st (t Td )
hR (t ) =
2 cos(w t +
n
)
)
wnTd ) p (t nTc Td )
n =
If Td = Td ,
y1 (t ) = st (t Td )hR (t )
2P cos(w t + w t + (t T ) + w T
2cos(w t + w T ) p(t nT T )
n d
o d
wiTd ) p (t iTc Td )
{2
y (t ) =
= 2 P cos( wo t + d (t Td ) woTd )
2.5.2
Noncoherent FHSS
2/27/2007
12/16
b)
c)
d)
BW = 2 L f = M f = Wd .
z For frequency synthesizer, there are k bits per chip, and 2 k frequency subbands. The
bandwidth of FH/MFSK is equal to BW = 2 k Wd = Ws
2/27/2007
13/16
2.5.3
z
ws
wd
Ts
Tc
wd
10
01
11
00
11
10
01
01
01
10
01
10
10
00
01
10
11
PN
0 0
Wd
Wd
WS
Wd
Wd
T
TS
TC
z Processing gain :
Jammer with power J
E
EW
J
If no FH, N J =
, b = b d
Wd N J
J
2/27/2007
14/16
E
EW
EW
J
, b' = b s = 2 k b d
Ws N J
J
J
Eb
Gp =
2.5.4
Eb
N J
= 2k
NJ
Tc < Ts , let Tc =
Ts
(N times frequency-hoping / symbol), E.g.,
N
a) L=2 (2 bits/symbol)
b) N=2 (2 FHs/symbol, or 1 FH/bit)
c) k=3 (there have 8 different frequencies/group)
2/27/2007
00
00
10
01
15/16
11
00
11
10
01
01
01
10
01
10
10
00
01
10
11
00
PN
Wd
Wd
WS
Wd
Wd
TC
T
TS
z The output is demodulated by using Noncoherent MFSK. Besides, each symbol could be
detected by majority decision rule among the N outputs (N FH outputs/symbol).
z Processing gain is equal to G p = 2 k
2.6
Hybrid DS/FH SS
z Widely used in military spread spectrum systems
z Transmitter
hT (t ) = 2 p (t nTc ) cos( wn t + n )
n =
2/27/2007
16/16
z Receiver
y1(t)
y2(t) y3(t)
st(t)
z If Td = Td = 0 ,
y1 (t ) = st (t )hT (t ) = c(t )
2 P cos( wo t + wn t + d (t ) + n ) p (t nTc )
2 p(t mT ) cos(w t +
c
m =
= c(t ) 2 P
{cos(w t + 2w t +
o
y2 (t ) = y1 (t ) |w = c(t ) 2 P
o
cos(w t +
o
2/27/2007