Update/Outline

• In this lecture we will consider

– Review Binary Digital Communications
– Derive Optimum Receiver Structure for
Generic Binary Modulations
– Popular Binary Modulation Schemes

1
 Binary Digital Data Transmission
• In digital systems, the signals take discrete
values to represent binary signals: -A, A is used to
represent 0 and 1, for example.

• The system model for baseband binary digital

data transmission is

Transmitter Σ Receiver
(+A,-A)
2
• Example of a digital signal and transmitted
waveform is

• Example of a received signal is

3
• For digital systems, the probability of error
is used as a measure of performance

• A possible receiver structure for detecting the

digital transmitted signals is shown below

V Threshold >0 choose +A

device
<0 choose -A
4
• The integrator averages out the noise received so
that the output waveform will look like

• Assume noise at the input to the receiver is AWGN

(Additive White Gaussian Noise)
5
 • Based on linear system analysis, the noise output
from the integrator will be Gaussian.
• Hence, we have

1
Σ V Decision

AWGN 6
With

7
 Problem - Error Probability

Prior Probability
Conditional Error Probability

8
 Error Probability Computation
• We have

• V is Gaussian with variance σ2 (which is the noise N variance).

• To compute Pe, we need to compute

P(E|0) and P(E|1) 9

With
10
• The key to estimating the error probabilities
is to find out more about the distribution
of the noise

• The noise mean can be calculated as

0

11
• Thus, we know that the output noise from the
integrator will have the following Gaussian
distribution

• Now,

13
• Due to symmetry, we have

• If “0” and “1” are equally likely, then

• Hence

14
 VIP Transformation

• Let

• where

Q[.] function
15
 Q-function
The Q-function is a standard form for expressing
error probabilities without closed form solution.

where erfc(.) is complementary

error function.

16
Plot

Approx

17
• Next note that one can represent as (“0”
sent) or (“1” sent).

A
T
t t
T
-A

bit “1” bit “0”

Energy
18
• We can use these energy calculations to find a
more general result for the error probabilities.
i.e.,

where

19
• Therefore,
~ SNR

• where
Complementary
Error Function

20
• A graph of Pe for baseband signaling is

Approx

21
 Optimum Receiver Structure
• So far we considered
a simple ad-hoc
receiver structure

• What is the optimum

receiver structure we
can have that will
minimize Pe

• The optimum
receiver structure is
the Matched Filter
22
 Linear Signal 01101
Σ VT
Processing
Unit

We could optimize

• Assuming that the noise is Gaussian after

signal processing (zero mean) so that
r.p.

where the “o” subscript stands for output

23
At ,

with

Assume “0” sent, then

and . Hence,

24
• Let , then

• Assume “1” sent, then

25
 Bit Error Probability

• Question: What is the optimal

threshold, ,which minimizes ?
• Solution: VT = arg min Pe (VT )
opt
V T

• Set
26
 Optimal Threshold

• For the important case where

we have

27
 Minimum Bit Error Probability

• As Z↑, Pe↓. Hence, to minimize Pe, one must maximize Z.

Thus, must find optimum signal processing unit represented
by a Filter, 28
Let

• To minimize Pe, one needs to maximize ρ.

⇒ By finding the “Optimal” or best filter
that maximizes ρ, we obtain the best
performance.
hopt (t) = arg max (h(t))
h(t)

• Given “0” transmitted →

• Given “1” transmitted →

29
 Optimal Receiver
1 Decision
VT
0

•
& Define

30
 1 Decision
VT
0

• Define

• Let . Then,

31
 ,

1 Decision
VT

0
• Let

• Instantaneous

32
• Problem: Find h(t) such that SNRI is
maximized at t = to. hopt (t) = arg max (h(t))
h(t)

Hopt (f ) = arg max (H(f ))

H(f )

• Solution:

33
 Schwarz’s Inequality

• Given and

with equality iff

where K is some constant.

34
Next let

Using Schwarz inequality we get

35
 ----Maximum SNR.

with Equality iff

36
 Optimal Receiver

with Equality iff

Note:
• Gaussian assumption made
• No White noise assumption made
37
 Optimum Receiver: Special Case
Assume a White noise input, then

Hence, where
and

38
• Since C = constant, ⇒ WLOG, the Optimal
Filter is the Matched Filter.

39
 Optimum (Matched filter) receiver for
binary signaling in white Gaussian noise

Decision:

+ t=T
Threshold
Σ comparison
-

2 Matched Filters (each matched to s1(t) and s2(t)) 40

Matched Filter

Matched Filter to s1(t)

41
 Correlator Receiver

• We know that the optimum receiver structure

consists of a matched filter so that

Matched Filter
42
 • We can also implement the matched filter
as a correlator as follows:

Correlator Receiver
Can easily prove that both receivers are equivalent (See Lecture
43

notes)
∴ Both filters are equivalent.
44
Optimum (Correlator) receiver for data
communications

+
Decision
Σ device
-

Both receivers are equivalent

45
Minimum Bit Error Probability

• Let

• Define

Correlation Coefficient
46
• But

• Hence,

• How shall we select the correlation coefficient

to minimize the error probability using the
optimum receiver structure?

47
 Popular Binary Modulation
Schemes

• Antipodal Signaling
• Non-Return to Zero (NRZ)
• Amplitude Shift Keying (ASK)
• Phase Shift Keying (PSK)
• Frequency Shift Keying (FSK)
• Differential PSK (DPSK)

48
Tx Bits: 1 0 1 1 0 1
Antipodal
baseband
signal

ASK

PSK

FSK
49
 Examples of Binary Digital
Modulation Schemes

• Antipodal Signaling

• Non-Return to Zero (NRZ)

50
• Amplitude Shift Keying (ASK)

51
 • Phase Shift Keying (PSK) or BPSK

VIP

Best Possible performance

52
• Frequency Shift Keying (FSK)

Let , and

assume that where n is an integer

Frequency separation

with Source data rate (bits/

sec)
Multiple of 53data rate
• Now,

0 Since

54
– Note:Can easily show that (actually, it
follows from the above). Hence, in this case,
and are ORTHOGONAL

55