Vector Concept of Vector Concept of Signals Signals
Vector Concept of SignalsVector Concept of Signals
Topics to be CoveredTopics to be CoveredVector space conceptRepresentation of signals using vectorsRepresentation of noise using vectorsRepresentation of modulation schemes
using vector space concept
2
Energy
dttsET
S )(0
2
Signal Space
Digitally modulated signal consists of selection / transmission of one out of a finite set of elements.The signals space concept is similar to vector space.Concept of vector space can be applied to represent digital modulated signals.It allows convenient visualization, mathematical modeling and comparison of various modulation schemes
],...,,[)( 21 nsssSts
Orthogonal Space of Basis Orthogonal Space of Basis FunctionsFunctionsAn N-dim orthogonal space consisting of linearly
independent basis functionsBasis functions are denoted byBasis functions must satisfy the following
conditions:
When Kj are nonzero, the signal space is called Orthogonal; When Kj=1, the signal space is called Orthonormal.
)(tj
Orthogonal Space of Basis Orthogonal Space of Basis FunctionsFunctionsAny finite set of waveforms can be expressed as
a linear combination of N orthogonal waveforms
Also called Generalized Fourier Series
Energy of a Signal Energy of a Signal
For orthonormal functions
Topics to be CoveredTopics to be CoveredMatched filterCorrelator receiversProbability of error
7
OPTIMUM RECEIVER FOR BINARYOPTIMUM RECEIVER FOR BINARY DIGITAL MODULATION SCHEMES DIGITAL MODULATION SCHEMES
The function of a receiver in a digital communication system is to distinguish between all the M transmitted signals sm(t), in the presence of noise
The performance of the receiver is usually measured in terms of the probability of error and the receiver is said to be optimum if it yields the minimum probability of error
4
m=1,2,…,M
Received Signal + Noise Vector Received Signal + Noise Vector
8
Demodulation of Digital SignalsDemodulation of Digital Signals
Matched FilterMatched FilterDetection of signal in presence of additive noiseReceiver knows what pulse shape it is looking for
Channel memory ignored (assumed compensated by other means, e.g. channel equalizer in receiver)
13 - 11
Additive white Gaussian noise (AWGN) with zero mean and
variance N0 /2
s(t)
Transmitted signal
w(t)
x(t) h(t) y(t)
t = T
y(T)
Matched filter
)()( )(*)()(*)()(
0 tntsthtwthtsty
T is the symbol period
Matched Filter DerivationMatched Filter DerivationDesign of matched filter
◦Maximizes signal power i.e. power at t = T◦Minimizes noise i.e. power of noise
Combined design criteria
13 - 12
power noise averagepower signal ousinstantane
)}({|)(|
SNR pulsepeak is where,max
2
20
tnETs
Matched Filter DerivationMatched Filter Derivation
13 - 13
dffHNdffGtnE N202 |)(|
2 )(} )( {
f2
0N
Noise power spectrum SW(f)
20 |)(|2
)(G )()( fHNffGfG HWN
)(*)()( thtwtn
AWGN Matchedfilter
PSDs are related by
Matched Filter DerivationMatched Filter Derivation
13 - 14
2 2 20 | )(S )(| |)(|
dfeffHTs Tfj
)()( )(0 fSfHfS
dfeffHts tfj )(S )( )( 2 0
)(*)()(0 thtsts
Find h(t) that maximizes pulse peak SNR
dffHN
dfeffH Tfj
20
2 2
|)(|2
| )(S )(|
Matched Filter DerivationMatched Filter Derivation
Schwartz’s inequality
For functions:
upper bound reached if
13 - 15
dffHN
dfeffH Tfj
20
2 2
|)(|2
| )(S )(|
Rkxkx )( )( *21
-
22
-
21
2
2-
1 )( )( )( )( dxxdxxdxxx
Matched Filter DerivationMatched Filter Derivation
13 - 16
)(s )( Hence,
inequality s' Schwartzby )(S )(
whenoccurs which , |)(| 2
|)(| 2 |)(|
2
| )(S )(
|)(| |)(| | )(S )(
)()( and )()(Let
2 *
2
0max
2
020
2 2
222 2
2 21
tTkth
efkfH
dffSN
dffSN
dffHN
dfeffH|
dffSdffHdfeffH|
efSffHf
opt
Tfjopt
-
Tfj
-
Tfj
Tfj
For real s(t)
Matched FilterMatched Filter Given transmitted signal s(t) of duration T, matched filter is
given by hopt(t) = k s(T-t) for any k Duration and shape of impulse response of the optimal filter
is determined by pulse shape s(t) hopt(t) is scaled, time-reversed, and shifted version of s(t) Optimal filter maximizes peak pulse SNR
SNR Does not depend on pulse shape s(t) Directly proportional to signal energy (energy per bit) E Inversely proportional to power spectral density of noise
13 - 17
SNR2|)(| 2 |)(| 2
0
2
0
2
0max
NEdtts
NdffS
Nb
Matched Filter ReceiverMatched Filter Receiver
13 - 18
Matched filter ExampleMatched filter ExampleReceived SNR is maximized at time T
T
example:
transmit filter receive filter
t
)()( tstTs
T t
)( ts
T t
)(ts
Matched Filter: optimal receive filter for maximizedNS
(matched)
Correlation Filter Correlation Filter
At t=T
At t=T, matched filter becomes a correlator
Correlator and Matched Filter Correlator and Matched Filter Outputs for a Sine Wave InputOutputs for a Sine Wave Input
Matched Filter ReceiverMatched Filter Receiver
13 - 22
Matched filter ExampleMatched filter ExampleReceived SNR is maximized at time T
T
example:
transmit filter receive filter
t
)()( tstTs
T t
)( ts
T t
)(ts
Matched Filter: optimal receive filter for maximizedNS
(matched)
Correlation Filter Correlation Filter
At t=T
At t=T, matched filter becomes a correlator
Correlator and Matched Filter Correlator and Matched Filter Outputs for a Sine Wave InputOutputs for a Sine Wave Input
Channel ModelChannel ModelAWGN channel model:
◦Signal vector is deterministic.◦Elements of noise vector are i.i.d
(independent and identically distributed) Gaussian random variables with zero-mean and variance . The noise vector pdf is
◦The elements of observed vector are independent Gaussian random variables. Its pdf is
26
),...,,( 21 iNiii aaas
),...,,( 21 Nrrrr
),...,,( 21 Nnnnn
nsr i
2/0N
0
2
2/10
exp1)(NN
pn
nn
0
2
2/10
exp1)|(NN
p ii
srsrr
Correlation Receiver Correlation Receiver for BPSK/BASKfor BPSK/BASK
One basis function, so one branchThe received signal is input to a correlator The correlator output is compared with a
threshold to decide which symbol was transmitted.
27
T
0
)(1 t)(tr Decision
CircuitCompare r
with threshold.
m̂
Sample at T
Probability of Error Probability of Error BPSK/BASKBPSK/BASK
29
a1(T) is the correlator output when s1(t) is senta2(T) is the correlator output when s2(t) is sentn0(T) is the correlator output when AWGN is present at the input
Output of the correlator:
Hypotheses: H1: s1 is assumed to have been transmitted H2: s2 is assumed to have been transmitted
Probability of Error:Probability of Error: BPSK/BASK BPSK/BASK
30
Probability of Error: BPSK/BASKProbability of Error: BPSK/BASK
31
Probability of ErrorProbability of Error BPSK/BASK BPSK/BASK
32
Probability of ErrorProbability of Error BPSK/BASK BPSK/BASK
33
Where Q(x) is the Complementary Error Function
And for BPSK
b
b
Ea
Ea
2
1
2/020 NNow
)2(0NEQP b
B
Coherent detection of BFSKCoherent detection of BFSK
For correct detection, the carriers must be orthogonal.
Error ProbabilityError Probability BPSK and BFSK BPSK and BFSK
BPSK and BFSK with coherent detection:
35
)(1 t
2s1sbE
“0” “1”
bE
)(2 t
bE221 ss
)(1 t
2s
1s
)(2 tbE
bEbE221 ss“0”
“1”
BPSK BFSK
2
0
NEQP b
B
2/
2/
0
21
NQPB
ss
0
NEQP b
B
Assignment#4Assignment#4Due Before start of next class on Due Before start of next class on
Saturday 23 April, 2011Saturday 23 April, 2011
◦ Q 1: Show that the probability of error for coherently detected BFSK is given by
◦ Q 2: Derive the value of decision threshold if the binary transmitted symbols in BPSK are not equiprobable.
◦ Hint: See Sec. 3.2 and App B.2.
36
0
NEQP b
B
Quiz#4Quiz#4◦Q : (a) A BASK signal transmits 2000 bps. If the
energy per bit is 4 Joules and the noise PSD is 1 W/Hz, how many erroneous bits will be received in 1 second?
◦(b) Also sketch the PDFs of the received signal for both the possible transmitted signals and indicate the area that corresponds to the probability of correct decision.
37
Coherent Detection Coherent Detection of MPSK of MPSK
38
Compute Choose smallest1
2arctanrr ̂
|ˆ| i
T
0
)(1 t
T
0
)(2 t)(tr
1r
2r
m̂
3s
7s
“110”)(1 t
4s 2ssE
“000”
)(2 t
6s 8s
1s5s
“001”“011”“010”
“101”
“111” “100”
8-PSK
Decision variable r̂z
Information resides in the phase, which needs to be detected
Coherent Detection of M-QAM Coherent Detection of M-QAM (Amplitude Phase Shift Keying)(Amplitude Phase Shift Keying)
Coherent detection of M-QAM
39
)(1 t
)(2 t
2s1s 3s 4s0000 0001 0011 0010
6s5s 7s 8s
10s9s 11s 12s
14s13s 15s 16s
1000 1001 1011 1010
1100 1101 1111 1110
0100 0101 0111 0110
16-QAM
©2000 Bijan Mobasseri 41
Signal space dimensionSignal space dimensionHow many basis functions does it take to
express a signal? It depends on the dimensionality of the signal
Some need just 1 some need an infinite number.
The number of dimensions is N and is always less than the number of signals in the set
N<=M
©2000 Bijan Mobasseri 42
Example: Fourier seriesExample: Fourier series
Remember Fouirer series? A signal was expanded as a linear sum of sines and cosines of different frequencies. Sounds familiar?
Sines and cosines are the basis functions and are in fact orthogonal to each other
cos 2nfot To cos 2mfot dt 0,m n
fo 1/To
©2000 Bijan Mobasseri 43
InterpretationInterpretations1(t) is now condensed into just two
numbers. We can “reconstruct” s1(t) like this
s1(t)=(1)1(t)+(-0.5)2(t)
Another way of looking at it is this
1
-0.5
1
2
©2000 Bijan Mobasseri 44
Signal constellationSignal constellationFinding individual components of each
signal along the two dimensions gets us the constellation
s4
s1
s2
s3
1
2
-0.5
-0.5 0.5
©2000 Bijan Mobasseri 45
Energy in simple languageEnergy in simple language
What we just saw says that the energy of a signal is simply the square of the length of its corresponding constellation vector
3
2 E=9+4=13E
©2000 Bijan Mobasseri 46
Constrained energy signalsConstrained energy signals
Let’s say you are under peak energy Ep constraint in your application. Just make sure all your signals are inside a circle of radius sqrt(Ep )
Ep
©2000 Bijan Mobasseri 47
Correlation of two signalsCorrelation of two signals
A very desirable situation in is to have signals that are mutually orthogonal. How do we test this? Find the angle between them
s1
s2 cos 12 s
1
Ts2
s1 s2
transpose
©2000 Bijan Mobasseri 48
Find the angle between sFind the angle between s11 and s and s22
Given that s1=(1,2)T and s2=(2,1)T, what is the angle between the two?
s1T s2 1 2
21
2 2 4
s1 1 4 5
s2 4 1 5
cos 12 4
5 5
45
12 36.9o
©2000 Bijan Mobasseri 49
Distance between two signalsDistance between two signals
The closer signals are together the more chances of detection error. Here is how we can find their separation
d12
2 s1 s22 s1 j s2 j 2
j1
N
(1)2 (1)2 2 d12 2
1 2
1
2
©2000 Bijan Mobasseri 50
Constellation building using Constellation building using correlator bankscorrelator banks
We can decompose the signal into its components as follows
s(t)
1
2
N
dt0
T
dt0
T
dt0
T
s1
s2
sN
N components
©2000 Bijan Mobasseri 51
Detection in the constellation Detection in the constellation spacespace
Received signal is put through the filter bank below and mapped to a point
s(t)
1
2
N
dt0
T
dt0
T
dt0
T
s1
s2
sN
componentsmapped to a single point
©2000 Bijan Mobasseri 52
Constellation recovery in noiseConstellation recovery in noise
Assume signal is contaminated with noise. All N components will also be affected. The original position of si(t) will be disturbed
©2000 Bijan Mobasseri 53
Actual exampleActual exampleHere is a 16-level constellation which is
reconstructed in the presence of noise
-2 -1.5 -1 -0.5 0 0.5 1 1.5 2-2
-1.5
-1
-0.5
0
0.5
1
1.5
2Eb/No=5 dB
©2000 Bijan Mobasseri 54
Detection in signal spaceDetection in signal space
One of the M allowable signals is transmitted, processed through the bank of correlators and mapped onto constellation question is based on what we see , what was the transmitted signal?
received signalwhich of the four did it come from
©2000 Bijan Mobasseri 55
Minimum distance decision ruleMinimum distance decision ruleIt can be shown that the optimum
decision, in the sense of lowest BER, is to pick the signal that is closest to the received vector. This is called maximum likelihood decision making
this is the most likely transmitted signal
received
©2000 Bijan Mobasseri 56
Defining decision regionsDefining decision regionsAn easy detection method, is to compute
“decision regions” offline. Here are a few examples
decide s1
decide s2
s1s2
measurement
decide s1decide s2
decide s3 decide s4
s1s2
s3 s4
decide s1
s1
©2000 Bijan Mobasseri 57
More formally...More formally...Partition the decision space into M
decision regions Zi, i=1,…,M. Let X be the measurement vector extracted from the received signal. Then
if XZi si was transmitted
©2000 Bijan Mobasseri 58
Error probabilityError probability
we can write an expression for error like thisP{error|si}=P{X does not lie in Zi|si was
transmitted}Generally
Pe P X Zi | si P{i1
M si}
©2000 Bijan Mobasseri 59
Example: BPSKExample: BPSK((BBinary inary PPhase hase SShift hift KKeying)eying)
BPSK is a well known digital modulation obtained by carrier modulating a polar NRZ signal. The rule is
1: s1=Acos(2πfct)0:s2= - Acos(2πfct)
1’s and 0’s are identified by 180 degree phase reversal at bit transitions
©2000 Bijan Mobasseri 60
Signal space for BPSKSignal space for BPSK
Look at s1 and s2. What is the basis function for them? Both signals can be uniquely written as a scalar multiple of a cosine. So a single cosine is the sole basis function. We have a 1-D constellation
A-Acos(2pifct)
©2000 Bijan Mobasseri 61
Bringing in EBringing in Ebb
We want each bit to have an energy Eb. Bits in BPSK are RF pulses of amplitude A and duration Tb. Their energy is A2Tb/2 . Therefore
Eb= A2Tb/2 --->A=sqrt(2Eb/Tb)We can write the two bits as follows
s1 t 2EbTb
cos 2fct
s2 t 2EbTb
cos 2fct
©2000 Bijan Mobasseri 62
BPSK basis functionBPSK basis functionAs a 1-D signal, there is one basis
function. We also know that basis functions must have unit energy. Using a normalization factor
E=1
1 t 2Tb
cos 2fct
©2000 Bijan Mobasseri 63
Formulating BERFormulating BER
BPSK constellation looks like this
√Eb-√Eb
X|1=[√Eb+n,n]
transmitted
received
noise
nois
e
Pe1 P Eb n 0 |1 is transmitted
if noise is negative enough, it will pushX to the left of the boundary, deciding 0instead
©2000 Bijan Mobasseri 64
Finding BERFinding BERLet’s rewrite BERBut n is gaussian with mean 0 and
variance No/2
Pe1 P Eb n 0 |1 P n Eb
-sqrt(Eb)
©2000 Bijan Mobasseri 65
BER for BPSKBER for BPSKUsing the trick to find the area under a
gaussian density(after normalization with respect to variance)
BER=Q[(2Eb/No)0.5]
©2000 Bijan Mobasseri 66
BPSK ExampleBPSK ExampleData is transmitted at Rb=106 b/s. Noise
PSD is 10-6 and pulses are rectangular with amplitude 0.2 volt. What is the BER?
First we need energy per bit, Eb. 1’s and 0’s are sent by
2Eb
Tbcos(2fct)
2EbTb
0.2
©2000 Bijan Mobasseri 67
Solving for ESolving for Ebb
Since bit rate is 106, bit length must be 1/Rb=10-6
Therefore, Eb=20x10-6=20 w-sec
©2000 Bijan Mobasseri 68
Solving for BERSolving for BERNoise PSD is No/2 =10-6. We know for BPSK
BER=0.5erfc[(Eb/No)0.5]What we have is then
Finish this using erf tablesBER
12erfc Eb
No
12erfc 2 10 7
2 10 6
12erfc( 0.1)
12erfc(0.316)
©2000 Bijan Mobasseri 69
M-ary signalingM-ary signalingBinary communications sends one of only 2 levels; 0
or 1There is another way: combine several bits into
symbols1 0 1 1 0 1 1 0 1 1 1 0 0 1 1
Combining two bits at a time gives rise to 4 symbols; a 4-ary signaling
©2000 Bijan Mobasseri 70
QPSK constellationQPSK constellation
45o
0001
11 10
√E
1 t 2T
cos2fct
2 t 2T
sin2fctBasis functions S=[0.7 √E,- 0.7 √E]
©2000 Bijan Mobasseri 71
QPSK decision regionsQPSK decision regions
0001
11 10
Decision regions re color-coded
©2000 Bijan Mobasseri 72
QPSK error rateQPSK error rate
Symbol error rate for QPSK is given by
This brings up the distinction between symbol error and bit error. They are not the same!
Pe erfc(E
2No
)
©2000 Bijan Mobasseri 73
Symbol errorSymbol errorSymbol error occurs when received vector is
assigned to the wrong partition in the constellation
When s1 is mistaken for s2, 00 is mistaken for 11
0011s1s2
©2000 Bijan Mobasseri 74
Symbol error vs. bit errorSymbol error vs. bit errorWhen a symbol error occurs, we might
suffer more than one bit error such as mistaking 00 for 11.
It is however unlikely to have more than one bit error when a symbol error occurs
10 10 11 1000
11 10 11 1000
10 symbols = 20 bits
Sym.error=1/10Bit error=1/20
©2000 Bijan Mobasseri 75
QPSK vs. BPSKQPSK vs. BPSK
Let’s compare the two based on BER and bandwidth
BER BandwidthBPSK QPSK BPSK QPSK
12erfc
EbNo
1
2erfc
EbNo
Rb Rb/2
EQUAL
©2000 Bijan Mobasseri 76
M-phase PSK (MPSK)M-phase PSK (MPSK)
If you combine 3 bits into one symbol, we have to realize 23=8 states. We can accomplish this with a single RF pulse taking 8 different phases 45o apart
si(t) 2ET
cos 2fct (i 1)4
, i 1,...,8
0
,0 t T
©2000 Bijan Mobasseri 77
8-PSK constellation8-PSK constellation
Distribute 8 phasors uniformly around a circle of radius √E
45o
Decision region
©2000 Bijan Mobasseri 78
Quadrature Amplitude Modulation Quadrature Amplitude Modulation (QAM)(QAM)
MPSK was a phase modulation scheme. All amplitudes are the same
QAM is described by a constellation consisting of combination of phase and amplitudes
The rule governing bits-to-symbols are the same, i.e. n bits are mapped to M=2n symbols
©2000 Bijan Mobasseri 79
16-QAM constellation using Gray 16-QAM constellation using Gray codingcoding
16-QAM has the following constellationNote gray coding where adjacent symbolsdiffer by only 1 bit
0010001100010000
1010
1110
0110
1011
1111
0111
1001
1101
0101
1000
1100
0100
©2000 Bijan Mobasseri 80
Vector representation Vector representation of 16-QAMof 16-QAM
There are 16 vectors, each defined by a pair of coordinates. The following 4x4 matrix describes the 16-QAM constellation
[ai ,bi ]
3,3 1,3 1,3 3,3 3,1 1,1 1,1 3,1
3, 1 1, 1 1, 1 3, 1 3, 3 1, 3 1, 3 3, 3
©2000 Bijan Mobasseri 81
What is energy per symbol in What is energy per symbol in QAM?QAM?
We had no trouble defining energy per symbol E for MPSK. For QAM, there is no single symbol energy. There are many
We therefore need to define average symbol energy Eavg
Eavg 1M
ai2 bi
2 i1
M
©2000 Bijan Mobasseri 82
EEavgavg for 16-QAM for 16-QAM
Using the [ai,bi] matrix and using E=ai^2+bi^2 we get one energy per signal
E
18 10 10 1810 2 2 1010 2 2 1018 10 10 18
Eavg=10
Relation between ERelation between Ebb/N/N00 & S/N & S/N
With R=Rb
83
In M-ary schemes
In most cases e.g. FSK
Time Division MultiplexingTime Division Multiplexing