Académique Documents
Professionnel Documents
Culture Documents
Abstract This document discusses the probability of error for non-coherent FSK and
DPSK. Digital signaling on a frequency selective channel will be reviewed as well as
equalization schemes to eliminate ISI.
Recall from the last lecture that we have Binary FSK orthogonal modulation. When a
1 is selected then S1(t) is transmitted and when a 0 is selected then S2(t)is transmitted.
S1(t) and S2(t) are orthogonal. The received signal is g1(t) if S1(t) was transmitted and
g2(t) if S2(t) is transmitted. The received signals g1(t) and g2(t) are also orthogonal. The
receiver structure is shown in figure 1.
Filter t=T
Matched to Envelope
l1
S1(t) Detector
l1 l2 1
X(t)
Compare
l2 l1 0
Filter Envelope l2
Matched to Detector
S2(t) t=T
l2 xI 2 xQ2
Where XI2 and XQ2 are both Gaussian with zero mean and psd = N0/2.
2l2 l2 2
exp , l2 0
f l2 (l2 ) N 0 N 0
0 , otherwise
1
Pr obability of error Pe1 Pr l2 l1 Pr l2 1 l1 f l1 l1 dl1
0
l 2
Pr l2 1 l1 f l2 l2 dl2 exp 1
l1 N 0
N N
where l1 xI1 xQ1 , xI1 : N E , 0 and xQ1 : N 0, 0
2 2
xI1 E
2
1
f xI xI1
N0
exp
N0
xQ
2
1
f xQ xQ1
N0
exp 1
N 0
Pr obability of error Pe1 p error xI1 xQ1 f xI xQ dxI1 dxQ1 1 1
2
xQ1 2 xI1 2 x I1 E
x Q1
2
E
2 2
Rewriting xQ1 xI1 xI1 E xQ1 2 xI1 E 2 xQ1
2 2 2
2
E
2
xQ1 2
Pe1
1
N0
exp
2 N
exp
2
N
x I1 E I1
dx exp dxQ1
N 0
0 0
1 E
Pe1
exp
2 2 N 0
2.0 Probability of error for non-coherent M-ary FSK
1 M i 1 kEb
1
i
Pe exp
2 m 1 i2 iN 0
Where k log 2 M
A non-coherent receiver for M-ary FSK is shown in Figure 2. As with the binary FSK
receiver a comparison of the output of the envelope detectors selects the appropriate
branch.
Figure 3 illustrates the probability of bit error for various values of M. As M increases
the bandwidth efficiency decreases but the power efficiency increases. More
constellations require more chunks of orthogonal spaces.
2
Filter t=T
Matched to Envelope
S1(t) Detector
t=T
Filter
Envelope
Matched to
X(t) Detector
S2(t)
Compare
t=T
Filter
Envelope
Matched to
Detector
Sm(t)
3
3.0 Differential Phase Shift Keying
Assumption: It works provided the unknown phase introduced by the channel varies
slowly (i.e. slow enough to be considered constant over 2 bit intervals). Consider input
sequence {mk}. Generate differentially encoded sequence {dk} from {mk} as follows:
dk = 1 =0
dk = 0 =
Example: if m k 10010011
d k 1 11011011
d k 10110111
0 00 000
4.0 - DQPSK
4
Exploit the above observation to derive the probability of error. Let the DPSK signal in 0
Eb
t Tb be cos 2 f c t . If the next bit (in the interval Tb t 2 Tb) is 1, then the
2Tb
phase is unchanged. If the next bit is 0, then the phase is shifted by .
4
Eb
cos 2 f c t , 0 t Tb
2Tb
S1 t
Eb
cos 2 f c t , Tb t 2T b
2Tb
Eb
cos 2 f c t , 0 t Tb
2Tb
S2 t
Eb
cos 2 f c t , Tb t 2T b
2Tb
1 E
Pe exp
2 N 0
Figure 4 compares the BER for DPSK and FSK. Observe that DPSK BER is
approximately 3 dB better than non-coherent FSK.
5
5.0 Digital Signaling on Frequency Selective Fading Channels
b t, x
x k xk ha t xk complex symbol sequence
]
amplitude sampling pulse
The above signal is transmitted through a channel c(t) that results in the received signal
w(t).
w(t ) xk h t kT z (t )
k 0
]
zero-mean AWGN
h(t ) h ( )c(t )d .
a
For causal channels h(t ) ha ( )c (t )d , t 0 .
0
We also assume h(t) = 0 for t 0 and h(t) = 0 for t LT. L is some real positive integer.
x (t nT )
n
u (t )
k
ha (t ) c (t ) h * (t ) y (n)
t T
z (t )
h (t )
Figure 5. Matched Filter Receiver in AWGN Channel
y (t ) x
k
k f (t kT ) v(t )
]
Filter noise
6
f (t ) h h( t )d .
*
f (t ) is the overall pulse response and it accounts for transient filter, channel and receive
filters.
yn y (nt ) x f (nT kT ) v(nT )
k
k
x f vk nk n
k
xn f 0 x
k
k f n 2 vn
k n
]
called intersymbol interference(ISI)
ISI is caused by the channel so if we can eliminate it then we can treat the channel as
AWGN channel. Therefore to achieve the same performance as on a AWGN channel, we
require x
k
k f n 2 0 . In order to do this we must make f = 0. Alternately,
n2
k n
0 if i j
f k k0 f 0 where ij .
1 if i j
Equally in the frequency domain
1 n
F f f0
T n T
, F ( f ) F f (t )
It can be shown that f(t) can be any function that has equally spaced zero crossings.
What is the optimum receiver? We must express w(t ) lim wn n (t ) using the n
Karhunen-Loeve expansion.
T
hnk h(t kT ) n* (t )dt
0
T
zn z (t ) n* (t )dt
0
N
1 1 2
p ( w x, H ) exp wn xk hnk
n 1 N 0 N 0 k
7
where H h1 , h 2 , h3 , ... ,hn and hn (...., hn 3 , hn 2 , hn 1 , h0 )
Therefore the optimum receiver is the Maximum Likelihood receiver, for the case of
AWGN it reduces to
N 2
arg max u ( x) wn xk hnk
x
n 1 k
y (t ) x
k
k f (t kT ) v(t )
is still Gaussian but is not white. Therefore the noise samples at the output are
correlated. In this case we can obtain a discrete time white noise model as
follows.
x (t nT )
n
k
ha (t ) c (t ) h * ( t ) Whitening
Filter
vk
t T
z (t )
h (t )
Figure 6. Whitening Filter with Matched Filter Receiver
where g n is the overall filter for the channel and the whitening filter and k is the
white noise process.
8
3. Another important consideration is the design of ISI equalizing filters is
extremely sensitive to timing information. We can solve this sensitivity in two
ways.
a. Use pulse shaping. Raised cosine pulse shaping allows us to derive the
length of the pulse by sampling at even points.
b. Use fractional sampling. Sample the output y(t) at a rate higher than 2/T.
Although we will still have correlated noise and will need a whitening
filter the overall pulse shape will be less sensitive to timing errors.
We consider the discrete time white noise model shown in figure 7. Where an is the
input signal, L is the memory of the channel, n is the additive noise and
g ( g 0 , g1, .....,gl )T is the channel vector that describes the overall channel impulse
response.
an T T ... T
g0 X g1 X gL X
rn g k an k n
k 0
n
Figure 7. Discrete Time White Noise Model
9
Linear equalizers estimate the nth transmitted symbol an as
M
The Zero Forcing Equalizer combines the channel and equalizer impulse response to
force to zero at all but one of the taps in the TDL filter. The tap coefficients, c, are chosen
to zero-out the ISI. Given the channel vector, g , we can select the tap coefficients, c,
to get the desired response q (0, 0, 0,......, qn , 0,...) . The Zero Forcing Equalizer zeros
out all but the desired result qn . The tap coefficients are calculated using the relationship
T
qn c g n . Zero forcing equalizers can perfectly eliminate ISI as M but it also
enhances the noise. In practice, the receiver does not know the channel vector and finite
length training sequences are used to choose the tap coefficients.
The MMSE is another symbol-by-symbol equalizer but is superior to the zero forcing
equalizer in performance. The MSME utilizes the mean square error criterion to adjust
the tap coefficients. We define an estimation error:
n an an
Where an is the symbol sent and an is the estimated symbol. The function to be
minimized is:
J min E n 2 min E
(an an ) 2
c c
M 2
min E (an c j rn j )
c
j M
The error is minimized by choosing c j so as to make the error sequence orthogonal to
the signal sequence rn l , for l M , i.e. E n rn l 0 , l M . The optimum c is
obtained by solving the following set of equations.
c E r
j M
j r
n j n l E an rn l , l M
10
@E rn j rnl , j l , M ,...., M
and b E an rn m E an rn m 1 ........E an rn m
then c opt 1 b
1 E n
2
c j (n 1) c j (n) , j M ....... M
2 c j
Where is a positive number.
Note:
E 2 n (n)
2 E ( n ) 2 E (n)rn j 2 Rer ( j )
c j c j
The term Rer is the cross correlation between the input and the error. The equalizer taps
can be obtained by implementing a stochastic gradient algorithm.
1
c j (n 1) c j (n) (n)rn j
2
To evaluate ( n ) in this algorithm we use a training sequence.
A DFE is a nonlinear equalizer that consists of a feed forward section and a feedback
section. DFEs are very effective in frequency selective channels because they mitigate
the effects of noise enhancements that degrade the performance of linear equalizers. The
DFE uses previous decisions to eliminate ISI caused by previously detected symbols on
current symbols. The output of the equalizer can be expressed as:
0 M2
an c j rn j
j M1
j 1
c j a%
n j
1 4 2 43 14 2 43
feed forwad filter M 1 1 taps feedback filter M 2 taps
Where a% %
n 1 , a %
n 2 , ....., an M 2 are earlier detected symbols. Both the feed forward loop
11
Recall the discrete-time white noise channel model.
L
rn g m an m n
m 0
Assume k symbols are transmitted over the channel. Then after receiving the sequence
rn n1 , the ML receiver decides in favor of the sequence an n 1 that maximizes the
k k
likelihood function
log rk , rk 1, ..., r1 ak , ak 1 ,..., a1
Since the noise samples are independent and rn depends only on the L most recent
transmitted symbols.
log rk , rk 1, ..., r1 ak , ak 1 ,..., a1 log rk ak , ak 1 ,., ak l log rk 1 , rk 2, .., r1 ak 1, ak 2 ,.., a1
1 4 4 4 4 4 2 4 4 4 4 43
This has already been calculated
where ak L 0 for k L 0
Since the 2nd term has been calculated at the previous time (k-1), only the first term needs
to be computed at time k for each incoming signal rk . The Veterbi algorithm can be used
to implement the MLSE equalizer. Adaptive MLSE is used in GSM.
References:
12