CSE4214 Digital Communications Baseband Demodulation/Detection

Chapter 3 Part 2
Baseband Demodulation/
Detection

CSE4214 Digital Communications Filtering Aspect of a Digital Communication System

Intersymbol Interference
n There are various filters throughout the system: at
transmitter, at receiver, and at channel.
n At the transmitter: transmitting filter Ht(f)
n At the channel, distributed reactances for cable or fading channel for
wireless. We use Hc(f) to represent the filtering effect.
n At the receiver: receiving filter Hr(f)
4

Page 1
1
Functions of Filters An Equivalent Model

1. Transmitting filter Ht(f): The input to the transmitting filter are impulses denoting the
information symbols {xk}. Bit 1 is represented by a positive impulse and bit 0 by a — Lumping all the filtering effects into one overall equivalent system
negative impulse. These impulses modulate rectangular pulses such that bit 1 is now transfer function: H(f) = Ht(f) Hc(f) Hr(f).
represented by a positive rectangular pulse and bit 0 by a negative rectangular pulse.
Since the bandwidth of a rectangular pulse is infinite, the rectangular pulses are band — The pulses at the output of the equivalent filter overlap each other.
limited by the transmitting filter. This effect is called intersymbol interference (ISI).
2. Channel Hc(f): The channel is modeled by a LTI system with a transfer function — ISI restricts the transmission rate of a communications system. Clearly
Hc(f). It accounts for distortion produced, for example, by fading or by bandwidth if the transmission rate Rs is increased (T is reduced), then the ISI will
constraints.
also increase.
3. Receiving filter Hr(f): The receiving filter compensates for distortion resulting from
the transmitter and the channel. It models both the receiving and equalizing filters. 5 6

Intersymbol Interference (ISI) ISI Example

n A form of distortion of a signal in which one symbol
interferes with subsequent symbols.
n Mainly caused by multipath propagation (in wireless)
or the inherent non-linear response of a channel.
n ISI causes successive symbols to “blur” together.
n ISI occurs even in the absence of noise (due to the
effects of filtering and channel-induced distortions)

7 8
In this interval: y(t) = b0 sB (t) + b1sB (t − T ) + b2 sB (t − 2T ) + b3sB (t − 3T ) + n0 (t)

Page 2
2
Nyquist Criterion for Zero ISI Activity 1
∞
y(t) = ∑bs k R (t − kT ) + n0 (t) Under what conditions the Nyquist Criterion for zero
k=−∞ ISI holds?
where sR (t) = hT (t)∗ hC (t)∗ hR (t) is the overall response of the
system due to a unit impulse at the input
#%
V if the k th bit is 1
bk = $
%& −V if the k th bit is 0
Normalize sR(0)=1, and take at sampling time t=mT
∞
y(mT ) = bm + ∑ bk sR (mT − kT ) + n0 (mT )
k=−∞
k≠m



ISI term

10

Zero ISI Example Nyquist Theoretical Minimum System Bandwidth

 The theoretical minimum bandwidth needed in order to detect Rs
symbols/s, without ISI, is Rs/2 Hz.
 This occurs when the system transfer function is made rectangular.
 For baseband systems, the rectangular bandwidth is called ideal Nyquist
filter.
 The sinc(t/T)-shaped pulse is called ideal Nyquist pulse.

Ideal Nyquist filter Ideal Nyquist pulse

11 12

Page 3
3
Raised Cosine Filter (1) Raised Cosine Filter (2)
 A system having the overall transfer function
H(f) as a rectangular pulse is difficult to be
 The impulse response of the
implemented: raised cosine filter is given by
1. The overall amplitude transfer function H(f)
has to be flat over the range -1/2T < f < 1/2T cos [2 π (W −W0 )t ]
and zero outside the range. This is physically h(t ) = 2W (sinc( 2W0t ) )
1−[4 (W −W0 )t ]2
unrealizable because the impulse response is
infinitely long and non-causal.  The minimum system
2. The synchronization of the clock in the
detector has to be perfect at instants t = kT. bandwidth required for a
 An alternative transfer function is the raised- symbol rate of Rs with the
cosine transfer function
raised cosine filter is given by
⎧1 for | f |< 2W0 − W
⎪
H ( f ) = ⎨cos 2 [
π f +W − 2W0
4 W −W0
] for 2W0 − W <| f |< W
W = 12 (1 + r ) Rs
⎪0 for | f |> W
⎩
where roll-off factor r = (W – W0)/W0
13 14

Zero ISI - Raised Cosine Example Activity 2

Find the minimum required bandwidth for the baseband transmission of a
4-level PAM pulse sequence having a date rate of R = 2400 bit/s if the
system transmission characteristic consist of a raised cosine spectrum with
100% excess bandwidth (r = 1).

15 16

Page 4
4
 CSE4214 Digital Communications Channel Characterization

— The baseband digital communication system is modified such that the equalizing and receiving
filters are considered as separate blocks.

Equalization

— The equivalent transfer function is obtained by lumping all of the transfer functions in one
system as
H(f) = Ht(f)Hc(f)Hr(f)He(f).
— In practice, the transmitting and receiving filters are chosen so that the product is equal to the
Nyquist raised cosine transfer function. For example,
Ht(f)Hr(f) = H (f).
RC

18

Equalization Equalizer Filter – Transversal Equalizer

x(k + n)

— The equalizing filter attempts to eliminate any distortion produced by the channel. If we model
the channel as z(k)
Hc(f) = |Hc(f)| ejθ (f)
c

where |H c (f)| represents the magnitude and θc(f) represents the phase. To compensate for the
channel distortion, the equalizing filter is implemented such as

1 1 — The transversal filter shown above is commonly used for equalization.

He ( f ) = = e− jθc ( f ) The output of the transversal filter is given by convolving the input samples and tap weights:
Hc ( f ) Hc ( f )
—

N
z[k ] = ∑ x[k − n]c
n=− N
n for k = −2 N ,…,2 N and n = − N ,…, N

19 20

Page 5
5
 Transversal Equalizer (2) Transversal Equalizer (3)
n The convolution can be expressed as: x(k + n)

⎡ z[−2 N ] ⎤ ⎡ x[− N ] 0 0 … 0 ⎤
⎢ z[−2 N + 1]⎥ ⎢ x[− N + 1] x[− N ] 0   ⎥
⎢  ⎥ ⎢    0 0 ⎥ ⎡c ⎤
⎢ z[−1] ⎥ ⎢ x[ N − 1] x[ N − 2] … x[− N ] 0 ⎥ ⎢ − N ⎥
⎢ 
⎢ z[0] ⎥ = x[ N ] x[ N − 1] … x[− N ] ⎥ ⎢ c0 ⎥ z(k)
⎢ ⎥ ⎢ ⎥ ⎢ ⎥
⎢ z[1] ⎥ ⎢ 0 x[ N ] … x[ − N + 1]⎥ ⎢  ⎥
⎢  ⎥ ⎢    ⎥ ⎣c
N ⎦

⎢ z[2 N − 1] ⎥ ⎢ 0 … 0 x[ N ] x[ N − 1] ⎥ c
⎢⎣ z[2 N ] ⎥⎦ ⎢⎣ 0 … 0 x [ N ] ⎥
⎦
      
z x
— Solving c = x-1z is possible only if x is a square matrix.
n Or a more compact form as: z=xc
— Alternatively, the zero-forcing solution is obtained by disposing the top N and bottom N rows of
n Solving c = x-1z is possible only if x is a square matrix. the matrix x and vector z.
n Note: the size of the vector z and the number of rows in the matrix x may be chosen to — To avoid ISI, the value of z is assumed to be
be any value. It depends on the number of ISI points interested
1 k =0
z[k ] = ⎧⎨
⎩ 0 k = ±1,..., ± N
21 22

Activity 5 Measuring ISI - Eye Pattern

 Eye patterns are obtained by applying the communication
system’s response to the vertical plates of an oscilloscope.
 A sawtooth wave with fundamental frequency equal to the
symbol rate is applied to the horizontal plates of the
oscilloscope.
 The setup superimposes the communication system’s
response within each signaling interval on top of each other
within a single interval (0,T).
1. A single impulse is transmitted through a digital communication system with the above  The maximum eye opening (MN) provides a good estimate of
waveform received, i.e. [x(k)]=[0.0, 0.0, 0.2, 0.9,-0.3, 0.1,0.0]. Use a zero-forcing solution to the optimal sampling instant.
find the weights {ck} (k=-1,0.1) of a 3-tap transversal equalizer that reduce the ISI so that the
 The width (ST) of the eye opening is the range over which the
equalized pulse samples z(k) have the values [z(-1)=0, z(0)=1, z(1)=0].
sampling may be performed.
2. Using the weights obtained in (1), determine the ISI values of the equalized pulse at sample
 The range (DA) of the amplitude variations provide an
times k = ±1, ±2, and ±3.
estimate of the ISI.
3. What is the largest magnitude sample contributing to the ISI and what is the sum of all of the
ISI magnitudes?  The range (JT) of the time differences of the zero crossings
provide an estimate of jitter.

23 24

Page 6
6