Inter Symbol

Interference
Communications Channels
 A physical medium is an inherent part of a
communications system
 Copper wires, radio medium, or optical fiber
 Communications system includes electronic or
optical devices that are part of the path followed by
a signal
 Equalizers, amplifiers, signal conditioners
 By communication channel we refer to the combined
end-to-end physical medium and attached devices
 Sometimes we use the term filter to refer to a
channel especially in the context of a specific
mathematical model for the channel
How good is a channel?
 Performance: What is the maximum reliable
transmission speed?
 Speed: Bit rate, R bps
 Reliability: Bit error rate, BER=10-k
 Focus of this section
 Cost: What is the cost of alternatives at a
given level of performance?
 Wired vs. wireless?
 Electronic vs. optical?
 Standard A vs. standard B?
Communications Channel
Transmitter Transmitted Received
Receiver
Signal Signal

Communication channel

Signal Bandwidth Transmission Impairments

 In order to transfer data  Signal attenuation
faster, a signal has to vary  Signal distortion
more quickly.
 Spurious noise
Channel Bandwidth
 Interference from other
 A channel or medium has signals
an inherent limit on how fast
 Limits accuracy of
the signals it passes can
vary measurements on received
signal
 Limits how tightly input
pulses can be packed
Frequency Domain Channel
Characterization
x(t)= Aincos 2ft y(t)=Aoutcos (2ft + (f))

Channel
t t
Aout
A(f) = Ain

 Apply sinusoidal input at frequency f

 Output is sinusoid at same frequency, but attenuated & phase-shifted
 Measure amplitude of output sinusoid (of same frequency f)
 Calculate amplitude response
 A(f) = ratio of output amplitude to input amplitude
 If A(f) ≈ 1, then input signal passes readily
 If A(f) ≈ 0, then input signal is blocked
 Bandwidth Wc is range of frequencies passed by channel
Ideal Low-Pass Filter
 Ideal filter: all sinusoids with frequency f<Wc are
passed without attenuation and delayed by t seconds;
sinusoids at other frequencies are blocked

y(t)=Aincos (2ft - 2ft )= Aincos (2f(t - t )) = x(t-t)

Amplitude Response Phase Response

1 (f) = -2ft
1/ 2
0
f

Wc f
Example: Low-Pass Filter
 Simplest non-ideal circuit that provides low-pass filtering
 Inputs at different frequencies are attenuated by different amounts
 Inputs at different frequencies are delayed by different amounts

Amplitude Response Phase Response

A(f) = 1 (f) = tan-1 2f
1
(1+42f2)1/2 1/ 2
0
f

-45o

-90o
f
Example: Bandpass Channel
Amplitude Response

A(f)

Wc f

 Some channels pass signals within a band that

excludes low frequencies
 Telephone modems, radio systems, …
 Channel bandwidth is the width of the frequency band
that passes non-negligible signal power
Channel Distortion
x(t) =  ak cos (2fkt + θk) Channel y(t)

 Let x(t) corresponds to a digital signal bearing data

information
 How well does y(t) follow x(t)?

y(t) = A(fk) ak cos (2fkt + θk + Φ(fk ))

 Channel has two effects:

 If amplitude response is not flat, then different frequency
components of x(t) will be transferred by different amounts
 If phase response is not flat, then different frequency
components of x(t) will be delayed by different amounts
 In either case, the shape of x(t) is altered
Example: Amplitude Distortion
x(t) 1 0 0 0 0 0 0 1
... ...

1 ms
t

 Let x(t) input to ideal lowpassfilter that has zero delay and
Wc = 1.5 kHz, 2.5 kHz, or 4.5 kHz

sin(  )cos(21000t)
4
x(t) = -0.5 +
 4
sin( 2 )cos(22000t) + sin(3 )cos(23000t) + …
4 4
+
 4  4
 Wc = 1.5 kHz passes only the first two terms
 Wc = 2.5 kHz passes the first three terms
 Wc = 4.5 kHz passes the first five terms
Amplitude Distortion
1.5 (a) 1 Harmonic
1
0.5  As the channel
0
bandwidth
0

1
0.5
0.25

0.75
0.125

0.375

0.625

0.875
-0.5
-1
-1.5
increases, the
output of the
1.5 (b) 2 Harmonics channel
1
0.5
resembles the
0
input more
0

1
0.5
0.25

0.75
0.125

0.375

0.625

0.875
-0.5
-1
-1.5
closely

(c) 4 Harmonics
1.5
1
0.5
0
0

1
0.5
0.25

0.75
0.125

0.375

0.625

0.875

-0.5
-1
-1.5
Time-domain Characterization
h(t)

Channel
t
0 t
td

 Time-domain characterization of a channel requires

finding the impulse response h(t)
 Apply a very narrow pulse to a channel and observe
the channel output
 h(t) typically a delayed pulse with ringing
 Interested in system designs with h(t) that can be
packed closely without interfering with each other
 Nyquist Pulse with Zero
Intersymbol Interference
 For channel with ideal lowpass amplitude response of
bandwidth Wc, the impulse response is a Nyquist pulse
h(t)=s(t – t), where T = 1/2 Wc, R = 2Wc and
s(t) = sin(2Wc t)/ 2Wct
1.2
1
0.8
0.6
0.4
0.2
0 t
-7T -6T -5T -4T -3T -2T T
-1-0.2 0 1T 2T 3T 4T 5T 6T 7T

-0.4

 s(t) has zero crossings at t = kT, k = +1, +2, …

 Pulses can be packed every T seconds with zero interference
 Example of composite waveform
Three Nyquist pulses +s(t) +s(t-T)
1
shown separately
 + s(t)

 + s(t-T) 0 t
-2 T -1T 0 1T 2T 3T 4T
 - s(t-2T)

Composite waveform
-1
r(t) = s(t)+s(t-T)-s(t-2T) -s(t-2T)
Samples at kT r(t) 2

r(0)=s(0)+s(-T)-s(-2T)=+1 1

r(T)=s(T)+s(0)-s(-T)=+1
0 t
r(2T)=s(2T)+s(T)-s(0)=-1 -2T -1T 0 1T 2T 3T 4T

Zero ISI at sampling -1

times kT -2
 Intersymbol Interference
 Intersymbol interference (ISI) occurs when a pulse spreads out
in such a way that it interferes with adjacent pulses at the sample
instant.

 Example: assume polar NRZ line code. The channel outputs are
shown as spreaded (width Tb becomes 2Tb) pulses shown
(Spreading due to bandlimited channel characteristics).

Channel Input Channel Output

Pulse width Tb Pulse width Tb

Data 1

 Tb 0 Tb  Tb 0 Tb

Data 0
 Tb 0 Tb  Tb 0 Tb
Intersymbol Interference
 For the input data stream:
1 0 1 1 0 1
A
0 Tb 2Tb 3Tb 4Tb 5Tb

 The channel output is the superposition of each bit’s

output:

0 Tb 2Tb 3Tb 4Tb 5Tb

Resultant
Channel Output
1 0 1 1 0 1 Waveform

0 Tb 2Tb 3Tb 4Tb 5Tb

 Intersymbol Interference
 If the rectangular multilevel pulses are filtered improperly as they pass through a
communications system, they will spread in time, and the pulse for each symbol
may be smeared into adjacent time slots and cause Intersymbol Interference.
 Intersymbol Interference
 Flat-topped multilevel input signal having pulse shape h(t) and values ak:
 
w in  t    an h  t  nTs    an h(t )*   t  nTs    an  t  nTs  * h t 
n n 
 t  1
Where h  t      Where D  pulses/s
 s
T T s

 
wout  t    an  t  nTs  * he  t    an he  t  nTs 
 n  n

Equivalent impulse response: h  t   h  t  * h t  * h t  * h t 

e T C R
 he(t) is the pulse shape that will appear at the output of the receiver
filter.
 Intersymbol Interference
 Equivalent Impulse Response he(t) :

h  t   h  t  * h  t  * h t  * h t 
e T C R
 Equivalent transfer function:
  t   sin  Ts f 
He  f   H  f  HT  f  H C  f  H R  f  Where H  f   F     Ts  
  s 
T   Ts f 
 Receiving filter can be designed to produce a needed He(f) in terms of HT(f) and HC(f):

He  f 
HR  f  
H  f  HT  f  H C  f 

 Output signal can be rewritten as:

wout  t    an he  t  nTs 
n
 He(f), chosen such to minimize ISI is called EQUALIZING FILTER)
Combating ISI
 Three strategies for eliminating ISI:
 Use a line code that is absolutely bandlimited.
 Would require Sinc pulse shape.
 Can’t actually do this (but can approximate).

 Use a line code that is zero during adjacent sample

instants.
 It’s okay for pulses to overlap somewhat, as long as there is no
overlap at the sample instants.
 Can come up with pulse shapes that don’t overlap during
adjacent sample instants.
 Raised-Cosine Rolloff pulse shaping

 Use a filter at the receiver to “undo” the distortion

introduced by the channel.
 Equalizer.
 Raised Cosine-Rolloff Nyquist Filtering
Consider other pulse shapes which require more bandwidth such
as the Raised Cosine-rolloff Nyquist filter but they are less affected
by synchrfonization errors.
 The Raised Cosine Nyquist filter is defined by its rollof factor
number r=fΔ/fo.
1, f  f1

 1     f  f1   
H e  f    1  cos    , f1  f B B is the Absolute Bandwidth
 2   2 f   

0, f B
f   B  f 0 f1  f 0  f  Where f o is the 6-dB bandwidth of the filter
f Rb
Rolloff factor: r  Bandwidth: B (1  r )
f0 2
 sin 2 f 0t   cos 2 f  t 
he  t   F  H e  f    2 f 0 
1
 
 2 f 0t  1   4 f  t  
2

f Rb
Rolloff factor: r  Bandwidth: B (1  r )
f0 2
 Raised Cosine-Rolloff Nyquist Filtering
 Now filtering requirements are relaxed because absolute bandwidth is
increased.
 Clock timing requirements are also relaxed.
 The r=0 case corresponds to the previous Minimum bandwidth case.
1     f  f1   
H e  f   1  cos  
2
  2 f   

B  fo  f

f R D
Rolloff factor: r  Bandwidth: B  (1  r )  (1  r )
f0 2 2
 Raised Cosine-Rolloff Nyquist Filtering
 Impulse response is given by:
 sin 2 f 0t   cos 2 f t 
he  t   F  H e  f    2 f 0 
1
 
 2 f 0t  1   4 f  t  
2

• The tails of he(t) are now

decreasing much faster than the Sa
function (As a function of t2).
• ISI due to synchronization errors
will be much lower.

Eeng 360 24
 Raised Cosine-Rolloff Nyquist Filtering

Frequency response and impulse

responses of Raised Cosine pulses for
various values of the roll off parameter.

rB
r   ISI 
 Raised Cosine-Rolloff Nyquist Filtering
 Illustrating the received bit stream of Raised Cosine pulse shaped
transmission corresponding to the binary stream of 1 0 0 1 0 for 3
different values of r=0, 0.5, 1.

1 0 0 1 0 1 0 0 1 0
Bandwidth for Raised Cosine Nyquist Filtering

 The bandwidth of a Raised-cosine (RC) rolloff pulse shape is a function

of the bit rate and the rolloff factor:
 f 
B  f o  f   f o 1     f o 1  r 
 fo 
R
B  1  r 
2
D
B  1  r  Multilevel Signalling
2
 Or solving for bit rate yields the expression:

2B
R
1 r
 This is the maximum transmitted bit rate when a RC-rolloff pulse
shape with Rolloff factor r is transmitted over a baseband channel with
bandwidth B.