Beruflich Dokumente
Kultur Dokumente
(ISI)
– it is a signal-dependent form of interference that arises
because of deviations in the frequency response of a
channel from the ideal channel.
– Example: Bandlimited channel
Time
Domain Bandlimited
channel
Frequency domain
BT.33
BT.34
1
Example
1 0 1
Waveform of ‘101’
BT.35
II. The received S/N ratio is high enough to ignore the effect of
channel noise (For example, a telephone system)
• control the shape of the received pulse.
BT.36
2
ISI
Consider a binary system, the incoming binary sequence
{bk } consists of symbols 1 and 0, each of duration Tb . The
pulse amplitude modulator modifies this binary sequence
into a new sequence of short pulses (approximating a unit
impulse), whose amplitude a k is represented in the polar
form
+ 1 if bk = 1
ak =
− 1 if bk = 0
ISI
∑ a δ (t − kT )
k
k b
{ak } :
Tb t
BT.38
3
ISI
The short pulses are applied to a transmit filter of impulse
response g(t), producing the transmitted signal
s (t ) = ∑ a k g (t − kTb )
k
ISI
The noisy signal x(t ) is then passed through a receive filter
of impulse response c(t ).The resulting output y (t ) is sampled
and reconstruced by means of a decision device.
x(t ) y(t) 1 if y > λ
Receive Decision
filter c(t ) device 0 if y < λ
Sample at ti = iTb
λ
The receiver output is y (t ) = µ ∑ a k p (t − kTb ) + n(t )
k
BT.40
4
ISI
Example: {bk } = 1101
{ak } : a1δ (t )
a2δ (t − Tb )
Tb t
y (t ) assume n(t ) = 0
t
µa1 p(t ) µa2 p (t − Tb ) y (t ) = µ ∑
k
a k p (t − kTb ) + n(t )
BT.41
ISI
The sampled output is
y (t i ) = µ ∑ a k p[(i − k )Tb ] + n(t i )
k
µ ∑ a k p[(i − k )Tb ] :
k
k ≠i
5
ISI
Example:{bk } = 1101 y (t i ) = µ a i + µ ∑a k p[(i − k )Tb ] + n(t i )
Tb k
k ≠i
y (t ) assume n(t ) = 0
t
µa2
y (t ) = µ ∑
k
a k p (t − kTb ) + n(t )
µa1 p (Tb ) (i = 2, k = 1)
µa3 p (−Tb ) = 0 (i = 2, k = 3)
t = t2 (i.e. i = 2) BT.43
Distortionless Transmission
In a digital transmission system, the frequency response of
the channel h(t ) is specified.
We need to determine the frequency responses of the
transmit g (t ) and receive filter c(t ) so as to reconstruct the
original binary data sequence {bk } .
{bk } Pulse- {ak } s (t ) xo (t ) x(t )
Transmit
amplitude Channel
filter g (t ) h(t )
modulator
w(t ) White noise
6
Distortionless Transmission
The decoding requires that
y (t i ) = µ ∑ a k p[(i − k )Tb ] + n(t i )
k Ignore the noise
0
= µai + µ ∑ a k p[(i − k )Tb ] + n(t i )
k
k ≠i
1 i = k
⇒ p (iTb − kTb ) =
0 i ≠ k
y (t ) y(ti) = µai
assume n(t ) = 0
t
BT.45
Distortionless Transmission
It can be shown that the condition
1 i = k
p (iTb − kTb ) =
0 i ≠ k
is equivalent to
∞
∑ P( f − n / T ) = T
n = −∞
b b
BT.46
7
Example
1 0 1
p (t )
p ( 0) p (Tb )
Sample points BT.47
Example
p( f ) 2
p (t )
sinc
Tb
1 / Tb
f
Tb 2Tb
∞
p( f ) p ( f − 1 / Tb ) p ( f − 2 / Tb ) ∑ p( f − n / T ) = T
b b
n = −∞
f
BT.48
8
Ideal Nyquist Channel
The simplest way of satisfying
∞
∑ P( f − n / T ) = T
n = −∞
b b
is a rectangular function:
1
−W < f < W
p ( f ) = 2W
0 | f |> W
1 / 2W
W = 1 / 2Tb
W = 1 / 2Tb
BT.49
sin( 2πWt )
p (t ) =
2πWt
9
Example
Sampling
instants
BT.51
BT.52
10
Raised Cosine Spectrum
We may overcome the practical difficulties encountered by
increasing the bandwidth of the filter.
1
−W < f < W W = 1 / 2Tb
Instead of using p ( f ) = 2W
0 | f |> W
we use
1
−W < f < W
p ( f ) + p ( f − 2W ) + P( f + 2W ) = 2W
0 | f |> W
BT.53
BT.54
11
Raised Cosine Spectrum
The frequency characteristic consists of a flat amplitude
portion and a roll-off portion that has a sinusoidal form. The
pulse spectrum p(f) is specified in terms of a roll off factor α
as follows:
1
0 ≤ f < f1
2W
1 π ( f − W
p( f ) = 1 − sin f1 ≤ f < 2W − f1
4W
2W − 2 f 1
0 f > 2W − f 1
The frequency parameter f1 and bandwidth W are related by
α = 1 − f1 / W
BT.55
BT.56
12
Raised Cosine Spectrum
The frequency response of α at 0, 0.5 and 1 are shown in
graph below. We observed that α at 1 and 0.5, the function
P(f) cutoff gradually as compared with the ideal Nyquist
channel and is therefore easier to implement in practice.
BT.57
BT.58
13
BT.59
Example
For α = 1, (f1 = 0) the system is known as the full-cosine
rolloff characteristic.
1 πf
1 + cos 0 < f < 2W
p ( f ) = 4W 2W
0 f > 2W
BT.60
14
Example
sinc( 2Wt )
p (t ) =
1 − 16W 2 t 2
BT.61
Example
This time response exhibits two interesting properties:
At t = ± Tb/2 = ± 1/4W we have p(t) = 0.5; that is, the
pulse width measured at half amplitude is exactly equal to
the bit duration Tb.
t = Tb/2
BT.62
15
Example
There are zero crossings at t = ± 3Tb/2, ± 5Tb/2, ... in
addition to the usual crossings at the sampling times t= ±
Tb/2, ± 2Tb/2,...
t = 3Tb/2
t = 5Tb/2
BT.63
Example
These two properties are extremely useful in extracting a
timing signal from the received signal for the purpose of
synchronization.
BT.64
16