Beruflich Dokumente
Kultur Dokumente
Original
message signal Source Channel
(analog) A/D Modulator
Encoder Encoder
Channel
Recovered
message Source Channel De-
D/A
signal Decoder Decoder modulator
(analog)
Ts
0 i t *j t dt i j
Ts N
s m, n s m t n* t dt s t sm,nn t
0 n 1
Ts N
Energy Es 2
sm t dt sm2 ,n sm 2
0 n1
sk t sk sk ,1 , sk , 2 ,..., sk , N k , l m 1,2,...M
sl t sl sl ,1 , sl , 2 ,..., sl , N
Ts N
1
dt = sk, sl =sk ×sl =å sk,ns*l,n
*
Correlation sk ( t) , sl ( t) =
Ts
òs ( t) s ( t)
k l
0 n=1
Ts N
Euclidean 2 2
d ( sk ( t) , sl ( t) ) = ò sk ( t) - sl ( t) dt =d ( sk, sl ) =å sk,n - sl,n
2 2
Distance 0 n=1
sm t Am cos 2f ct Re Am e j 2fct sm,lp t Am
2 Ts cos 2f ct , 0 t Ts
t
0, otherwise
Signal-Space (Vector Space) Representation (Obtained through the use of
basis functions)
sm t sm Am Ts 2 1-dimensional
Ts
Signal Energy Esm sm2 t dt Am2 Ts 2
0
ECE414 Wireless Communications, University of Waterloo, Winter 2012 9
M-ASK (cont’d)
sm t A cos 2f ct m Re Ae j m e j 2fct sm,lp t Ae j m
Basis Functions
2 Ts cos 2f ct , 0 t Ts 2 Ts sin 2f c t , 0 t Ts
1 t 2 t
0, otherwise 0, otherwise
Signal-Space Representation
s1 Es ,0
1 0 1 1 0
A
s 2 Es ,0 t
-A
Signal-Space 2 t
Representation Es
Es Es
1 t
Es
Examples of QAM
Signal Constellations
Basis Functions
2 Ts cos 2f ct , 0 t Ts 2 Ts sin 2f c t , 0 t Ts
1 t 2 t
0, otherwise 0, otherwise
Signal-Space Representation
Signal Energy
Ts
E sm sm2 t dt Am2 ,r Am2 ,i Ts 2
0
sm t A cos 2 f c f m t Re Ae j 2fmt e j 2fct sm,lp t Ae j 2fmt
Cross Correlation
sin 2T k l f
k ,l
2T k l f
For f 1 2T and k l k ,l 0
Therefore, the minimum frequency separation between adjacent signals for
orthogonality of the M signals is f 1 2T
ECE414 Wireless Communications, University of Waterloo, Winter 2012 16
M-FSK (cont’d)
Assuming frequency separation f 1 2T , the signal-space representation
for the M-FSK signals are given as N-dimensional vectors, where N=M.
2 Ts cos 2 f c f m t , 0 t Ts
m t
0, otherwise
s1 t s1 Es 0 0 ...... 0
s2 t s 2 0 Es 0 ...... 0
.
.
.
sM t s M 0 0 0 ...... Es
Ts
where Esm Es sm t dt A Ts 2
2 2
0
Temporal efficiency: How wide are the time variations of the transmitted signal?
Temporal efficiency=Peak power/Average power
The choice of amplifier depends on the temporal characteristics of the signal.
Other considerations:
• Hardware/software implementation complexity & cost of implementation
• Sensitivity to interference
• Robustness to impairments encountered in a wireless channel
g t ak p t kTs ak : Baseband modulation symbol
k Ts : Signal interval p t : Pulse shape
1
g f P f a f
2
where P f F p t
Ts
a f Ra n e j 2fnTs
See Ch.4 of Digital Communications
1 n é * ù
by Proakis for the proof Ra ( n) = E ëakak+nû
2
ECE414 Wireless Communications, University of Waterloo, Winter 2012 22
Example: PSD of BPSK with Rectangle Pulse Shaping
Baseband equivalent of BPSK sequence
g t ak p t kTs ak A Independent data symbols are assumed
k
Ra n E ak ak* n
E ak2 , n 0 A2 , n 0
E ak E ak* n , n 0 0, n 0
a f F Ra n Ra n e j 2fnTs A2
n
Pulse shaping
æt- Ts / 2 ö FT
p(t) p( t) =Õ
- j 2 p f ( Ts 2)
ç ÷¬ ¾ ® P ( f ) =Tssinc ( fTs ) e
è Ts ø
t P ( f ) =Tssinc ( fTs )
T/2 T
S f G f f c G * f f c See
s t Re g t e j 2fct
FT 1
2 Tutorial 1
PSD of bandpass BPSK sequence
s f
1
4
g f f c *g f f c
See Ch.4 of Digital Communications
by Proakis for the proof
1 1
A2Ts sinc 2 f f c Ts A2Ts sinc 2 f f c Ts
4 4
Null-to-null bandwidth
E ak jbk 2 , n 0 10 A2 , n 0
E ak jbk ak n jbk n , n 0 0, n0
PSD of baseband QAM sequence Note that PSD of QAM has the
same general form as BPSK.
g f 10 A2Tsinc 2 fT
Half-Sinusoid Pulse
t
p t A sin , 0 t T
T
AT jfT 1 1
P f e sinc fT - sinc fT
2 2 2
Full-Cosine Pulse
A t
p t 1 cos 0 t T
2 T
AT jfT
P f e 2sinc fT sinc fT 1 sinc fT 1
4
1
T , 0 f
2T
T T 1 1 1
P f 1 cos f , f
2
2 T 2T 2T
1
0 , f
2T
ECE414 Wireless Communications, University of Waterloo, Winter 2012 28
Comparison of Pulse Shapes
Time-Domain
Square
Full-cosine
Half-sinusoid
Gaussian
Full-cosine • Square
Half-sinusoid
BW=2/T
Gaussian
• Half-sinusoid
Square BW=3/T
• Full-cosine
BW=4/T
2/T
3/T
4/T
ECE414 Wireless Communications, University of Waterloo, Winter 2012 30
Comparison of Pulse Shapes (cont’d)
Raised Cosine
0 1
α: Roll-off factor
1
BW
T
1 2
BW
T T
1/T
2/T
ECE414 Wireless Communications, University of Waterloo, Winter 2012 31
Optimum Receiver for AWGN
For a given SNR (i.e. a given signal power for fixed noise power), we aim to
achieve a low error probability. To calculate error probability, first we need to
identify the receiver structure.
The receiver consists of a demodulator and a detector:
• The demodulator converts the received waveform r(t) into a N dimensional
vector r r1 , r2 ,...rN where N is the dimension of the signal-space for the
given modulation type.
• The detector decides which of the possible M signal waveforms was
transmitted based on r, where M is the constellation size.
sm t r t r ŝm
Demodulator Detector
n t r t sm t n t r s m n
This decision criterion is called the Maximum A Posteriori Probability (MAP) rule.
max p s m r
m1, 2...M
Bayes Theorem
p r s m p s m
max
m1, 2...M p r p r : Common for all
max p r s m p s m
m1, 2...M p s m 1 M, i.e. Equally probable
max p r s m messages
m1, 2...M
The conditional pdf p r s m is called the likelihood function and the decision
criterion based on the maximization of p r s m over the M signals is called the
maximum likelihood (ML) criterion.
ECE414 Wireless Communications, University of Waterloo, Winter 2012 34
Optimum Receiver for AWGN (cont’d)
r sm n
For an AWGN channel, the components of the noise vector n are zero-
mean Gaussian random variables with variance N0/2
1 nk2 1 nk2
f nk exp 2 exp
k 1,2...N
2 2
2 2 N0 2
N 0 N0
p r s m p rk sm,k f rk sm,k
N N
k 1 k 1
1 1 2
rk sm,k
N
exp
k 1 N 0 N0
1 1 N 2
exp rk sm,k
N 0 N 2
N 0 k 1
Unlike other M-PSK for M>2, we can represent this special form of BPSK
signal as 1-dimensional signal. The basis function is given as
2 T cos 2f c t , 0 t T
1 t
0, otherwise
Therefore, the optimal receiver has the following form of
r t T r Euclidean
.dt Distance
0 Decoder
1 t
ECE414 Wireless Communications, University of Waterloo, Winter 2012 37
Example: Error Probability for BPSK (cont’d)
Assume s1(t) is sent. Under the assumption of AWGN, the received signal
r t s1 t w t
where
def T
n w t 1 t dt ~ N 0, N 0 2
0
E E
P bˆ 0 b 1 f r b 1dr
0
1 r E 2
exp dr
N 0 0 N 0 r E
y
N0 2
1 y2
exp dy
2 2 E N0 2
2E
Q
N0
1 y2 2
where Q-function is defined as Q x e dy
2 x
T
.dt Detector
0
r t 1 t sˆ min r sm
2
m1, 2,3, 4
T
.dt
0
2 t
First, we calculate P(c), i.e. the probability of making a correct decision. Then,
probability of error is simply found as P(e)=1-P(c).
ECE414 Wireless Communications, University of Waterloo, Winter 2012 42
Example: Error Probability for QPSK (cont’d)
Assume that the signal located at the origin has been transmitted. If the
received signal is in the shaded area, this means we will make a correct
decision.
def T
d 2 Es nI w t 1 t dt ~ N 0, N 0 2
d 2
d 0
def T
d 2 nQ w t 2 t dt ~ N 0, N 0 2
0
P c s1 P nI d 2, nQ d 2
P nI d 2 P nQ d 2
P n Q
2
Q
Es
n ~ N , 2
N0
Q 1 Q
2
Es
1 Q
N 0
Due to symmetry,
P c P c s1 P c s2 P c s3 P c s4
2
Es Es
P e 1 P c 2Q Q
N 0 N 0
Es 2 Eb
2
2 Eb 2 Eb
2Q Q
N0 N0
T
.dt Detector
0
r t 1 t sˆ min r sm
2
m1, 2,3, 4
T
.dt
0
2 t
2 Ts cos 2 f c 1 2T t , 0 t Ts
1 t
0, otherwise
2 Ts cos 2 f c 1 T t , 0 t Ts
2 t
0, otherwise
ECE414 Wireless Communications, University of Waterloo, Winter 2012 45
Example: Error Probability for BFSK (cont’d)
2 t
d 2 Es By rotation, it can be easily shown that
Es
Es
P e Q
Es
1 t N0
Es
Due to symmetry, P e P e s1 Q
N0
ECE414 Wireless Communications, University of Waterloo, Winter 2012 46
A Union Bound on Error Probability
In most cases, probability of error can not be obtained in closed form.
Therefore, one needs to find some bounds or approximations which can work for
any signal constellation.
We have already shown that the optimal decoder for any signal constellation
2
over AWGN is given by the Euclidean distance decoder, i.e. d m r s m
M 1 M
P e P e s m P s m P e s m P e s m : Probability of making a
m1 M m1 decision error when sm was sent
M
P e s m P dl d m s m
Union Bound (U-B)
l 1
l m
M
P Ai P Ai
P d l d m s m i i
l 1
l l P s m s l : The probability of choosing sl
M
P s m sl s m instead of the originally transmitted sm
l 1
l l
1 M 1 M M d l ,m
P e P e s m Q
M m1 M m1 l 1 2 N 0
l m
The U-B requires the computation of all distances dl,m among signals in
the constellation. A looser bound can be obtained as follows
U B M d d “Minimum Euclidean
P e s m Q l ,m M 1 Q min
l 1 2 N 0 2N0 distance” bound
l m
1 M 1 M M dl ,m M 1 Q d min
P e P e s m Q 2N
M m1 M m1 l 1 2 N 0 0
l m
1 M ~ 1 M d min
P e P e s m N dmin,m Q
M m1 M m1 2N0 1 M
N dmin N dmin,m
M m1
d min
N dmin Q
2N0
2
d min 4 Es sin 2 4 Eb log 2 M sin 2
Es M M
1, M 2
N dmin
2, M 2
2
Replacing d min and N dmin into the formula on p.50, we obtain
2 Eb
Q , M 2
N 0
P e
2Q 2 Eb log M sin 2 , M 2
N 2
0 M
2 8
d min 2 A Ts 2 Esavg Ebavg
5 Es 4 Eb
5
~ d min 4 Ebavg
P e N dmin Q 3Q
2N0 5 N0
Es Eb log 2 M
P e M 1 Q
M 1 Q
N0 N0
• BPSK
2 Eb
P e Q
N0
• BFSK
Eb
P e Q
N0
d min
2
P 10 log10
4 Eb
dk 0 1 1 1 0 1
bk +1 -1 -1 -1 +1 -1
ak +1 +1 -1 +1 -1 -1 +1
A cos c t
θ represents any mismatch between transmitter/receiver oscillators or phase
introduced by the channel. In our system model, (independent of where it
comes from) we included in the transmitter block.
In coherent systems, we need to estimate and compensate this phase error at
the receiver. Here, we simply ignore it!
yI t
dt
r t t kT yk zk
2 cos c t sgn Re .
yQ t
y k* 1
dt
1 Symbol
Delay *
2 sin c t
ECE414 Wireless Communications, University of Waterloo, Winter 2012 60
Error Probability of DPSK
y t y I t jyQ t
ak Ae j N t
def def
Defining 1 y k y k 1 2 2 y k y k 1 2 the decision variable
can be written as
2 2
z k sgn Re y k y k* 1 1 2
1 ATe j ak ak 1 / 2 N k N k 1 2
2 ATe j ak ak 1 / 2 N k N k 1 2
E 1 ATe j ak ak 1 / 2
E 2 ATe j ak ak 1 / 2
Var 1 E 1 E 1 E 14 N N
2
k k 1 N k N k 1 * N 0T
Var E E E N N
2 1 *
2 2 2 k k 1 N k N k 1 N 0T
4
ak ak 1 bk ak ak 1 2 ak ak 1 2
+1 +1 +1 +1 0
-1 +1 -1 0 -1
-1 -1 +1 -1 0
+1 -1 -1 0 1
E 2 ATe j
2 R ~ N AT cos , 2 2 I ~ N AT sin ,
2
Var 2 N 0T 2 2 R j 2 I Complex Gaussian
1 : Rayleigh, 2 : Rician
ECE414 Wireless Communications, University of Waterloo, Winter 2012 63
Error Probability of DPSK (cont’d)
1 1 2
1 : Rayleigh f 1 2 exp
2 2
2 : Rician
1
f 2 22 exp 2 2 2
2
2
I
0 2
2
2 2 2 2 2
where the non-zero mean is found as A T cos sin AT
Now, we return to P(e)
computation
P e P 1 2 P 1 2 f 2 d 1 : Rayleigh
0
2 : Rician
1
P 1 2 2
1
exp 2 d 1
2
def 1
u
2 2
exp u du
2 2 2
exp 2 2 2
ECE414 Wireless Communications, University of Waterloo, Winter 2012 64
Error Probability of DPSK (cont’d)
2 2 2
P e exp 2 2 exp I
2 0 2
d
0 2 2 Variable change
def
1 x x
2 2 x x 2
exp I0
2
2 0 2 2 2 2 dx
def
1 m2 x x 2 m 2 mx m 2
exp 2 2 exp I
2 0 2
dx AT
2 2 0 2
=1
1 A T 2 2 N 0T 2
exp
2 2N0 E s 2 t dt A 2T 2
1 E T
exp
2 N0
-1
10
-2
10
BER
-3
10
-4
10
-5
10
-6
10
0 2 4 6 8 10 12
SNR [dB]
w t
“Actual” Channel
“Equivalent” Channel
ISI terms
The condition for no ISI is
1, n 0
heq nTs
0, otherwise
In frequency domain, this requires
l
H eq f constant
l Ts
See proof Proakis “Digital Communications” Chapter 9
This condition is known as “Nyquist pulse-shaping criterion” or “Nyquist
condition for zero ISI”.
ECE414 Wireless Communications, University of Waterloo, Winter 2012 68
Temporal Characteristics of Modulation Schemes (cont’d)
In the following, we consider three distinct cases:
W: Bandwidth of equivalent ch.
1 l
H eq f
Ts l Ts
2W
Ts
2W
l
H eq f
1 l Ts
Ts
2W
l
For this case, there exists many solutions as to satisfy H eq f cons.
l Ts
A particular pulse shape which satisfies the above property and has been widely
used in practical applications is “raised cosine”. (See page 28) The “Nyquist” pulse
takes zero at the sampling points for adjacent signalling intervals.
X f l Ts cons.
l
ECE414 Wireless Communications, University of Waterloo, Winter 2012 70
Temporal Characteristics of Modulation Schemes (cont’d)
Under the matched-filter assumption (i.e. which maximizes the output
signal-to-noise ratio), the transmit and receive filters satisfy
HT f H R f
Under the ideal channel assumption , i.e. H C f 1
HT f H R f H eq f
• Minimum instantaneous
power=0
• Maximum instantaneous
power=(1.6)2=4.1 [dB]
• Dynamic range=4.1 [dB]
• Average power=1=0 [dB]
• Peak power Avg. power 4.1 dB
+ =
t
s t cos2f c t 2h g d cos 2f c t 2h ak q t kTs
k
where
def t
q t p kTs d h: Modulation index
n 1 t nTs han
t ; a 2h ak 2h an 2 t n hna n nTs t n 1 Ts
k 0 2T s 2Ts
n
han
s t cos 2f ct t ; a cos2 f c t n hnan
2Ts
h
an 1 s t cos 2 f c t n hn
2Ts
h
an 1 s t cos 2
c f t hn
2Ts
n
For orthogonality, the minimum value for h should be chosen as h=1/2. This
special case is known as “Minimum Shift Keying” (MSK).
ECE414 Wireless Communications, University of Waterloo, Winter 2012 78
Continuous FSK (cont’d)
We have already introduced MSK as a special case of modulation family of
CFSK.
An MSK signal can be also considered as a special form of OQPSK where
the rectangular pulses are replaced with half-sinusoidal pulses.
t
cos ,0 t 2Ts
p t 2Ts
0, otherwise
The transmission rate on the two orthogonal carriers is 1/2Ts bits/sec so that
the combined transmission rate is 1/Ts bits/sec.
Continuous phase is assured in MSK while 90 and 180 phase changes are
observable for OQPSK and QPSK respectively.
ECE414 Wireless Communications, University of Waterloo, Winter 2012 80
Comparison of MSK, QPSK and OQPSK (cont’d)
In terms of temporal
efficiency, MSK obviously
outperforms QPSK and
OQPSK.
The main lobe of MSK is
wider than that of QPSK and
OQPSK and, in terms of null-
to-null bandwidth MSK is less
spectral efficient.
MSK has lower sidelobes
than QPSK and OQPSK
Less adjacent channel
interference
MSK, QPSK and OQPSK
have the same power efficiency.
ECE414 Wireless Communications, University of Waterloo, Winter 2012 81
Gaussian MSK
The spectral efficiency of MSK can be further improved by prefiltering.
f 2 ln 2 2 2 2 B 2t 2
H f exp h t
B exp
B 2 ln 2 ln 2
t T / 2 t T /2
x1 x2
T T
Phase pulse corresponding to rectangular
pulse shaping (i.e. no filtering) is also
included in the figure.
ECE414 Wireless Communications, University of Waterloo, Winter 2012 83
Gaussian MSK (cont’d)
BT: Normalized
3dB-Bandwidth
of Gaussian filter
For BT ∞, the pulse shape takes its original “unfiltered” form , i.e. rectangle
pulse. GMSKMSK
The frequency pulse has a duration of 2Ts although signaling rate is 1/Ts. Such
a LPF will result in intersymbol interference which requires sequence estimation
for optimal detection.