tele3113 wk11tue
DESCRIPTION
TRANSCRIPT
6 Oct. 2009 TELE3113 1
TELE3113 Analogue and Digital Communications –Detection Theory
School of Electrical Engineering and TelecommunicationsThe University of New South Wales
6 Oct. 2009 TELE3113 2
Digital Signal Detection
Receive filter
n(t)AWGN
si(t)r(t)
Decision device
Sampled at t=kTs
1 if y(kTs)>λ
y(t) y(kTs)
Thresholdλ
0 if y(kTs)<λ
At the receiving end of the digital communication system:
Noise power spectral density: Sn(ω)=η/2
≤≤−=≤≤+=
=0for 0 )(1for 0 )(
)(2
1
TtAtsTtAts
tsi
Polar NRZ Signaling
6 Oct. 2009 TELE3113 3
Suppose there are M possible signal symbols: {si} for i=1,…,M
We can represent these symbols in vector form
Similarly the noise vectors, and the received signal vectors,
Thus
Digital Signal Detection
rrnr
nsr iirrr
+=
ϕ1
ϕ2
ϕ3
1sr
3sr 2sr
4sr1rr
nr
nr
nr
nr
isr
6 Oct. 2009 TELE3113 4
Digital Signal DetectionIn each time interval, the signal detector makes a decision based on the observation of the vector so that the probability of correct decision is maximized.
rr
Consider a decision rule based on the posterior probabilities
)|21for ) vector received| ted transmit was signal(
rsP( ,M,, i rsP
i
irr
Krr
==
The decision is based on selecting the signal corresponding to the maximum set of posterior probabilities.
)(
)()|()|
rfsPsrf
rsP( iii r
rrrrr=
where is probability of si being transmitted and is the pdf function of the received signal vector .
This kind of decision is called maximum a posteriori probability (MAP) criterion
)rf(r
rr
Choose to maximize:
∑=
=M
1mms|rf()rf( where )() msP rrrrisr
)( isP r
6 Oct. 2009 TELE3113 5
If the M symbols are equally probable; i.e. for all i, the decision rule based on finding the signal that maximizes is equivalent to finding the signal that maximizes .
Digital Signal Detection
The decision based on the maximum of over M signal symbols is called the maximum-likelihood (ML) criterion
)| rsP( irr
)s|rf( irr
∑=
==M
1mms|rf()rf( where )()
)()()|(
)| mii
i sPrf
sPsrfrsP( rrrr
r
rrrrr
MAP criterion
)s|rf( irr
where is called the likelihood function.)s|rf( irr
MsP i /1)( =r
6 Oct. 2009 TELE3113 6
Thus {rk} are statistically independent Gaussian variables
where N is number of base vectors
and
Digital Signal Detectionnsr irrr
+=Recall:
For AWGN, the noise {nk} components are uncorrelated Gaussian variables which are statistically independent
ikkikkk s]nE[s]E[r , ]E[n =+== mean) (zero 0
( )22
][][ 22222 ησσησ ==→=−= nrn nEnEVariance
)s|f(r)s|rf( ikki
N
k 1=Π=
rr
η
σ
πη
σπ
/)(
)2/()(
2
22
ikk
nikk
sr
sr
n
e
e
−−
−−
=
=
1
21)s|f(r ikk
Take natural logarithm on both sides, gives
( ) ( )∑=
−−−
=N
kikk srN
1
21ln2
lnη
πη)s|rf( irr
6 Oct. 2009 TELE3113 7
Digital Signal Detection( ) ( )∑
=
−−−
=N
kikk srN
1
21ln2
lnη
πη)s|rf( irr
With ln(•) is a monotonic function, the maximum of over is equivalent to finding the signal that minimizes the Euclidean distance:
)s|rf( irr
isr
isr
∑=
−=N
kikki srsrD
1
2)(),( rr
So, the ML decision criterion (maximize over M signal symbols, i.e. i=1,…M) reduces to finding the signal that is the closest in distance to the received signal vector .
isr)s|rf( i
rr
rr
Example: 3 signal symbols.
Note the decision regions formed by the perpendicular bisectors of any two signal symbols.
6 Oct. 2009 TELE3113 8
Digital Signal DetectionDetection error will occur when the received signal vector falls into the decision region of other signal symbols. This is due to the presence of strong random noise.
Consider there are two signal symbols s1 and s2 , which are spaced d apart. The decision boundary is their perpendicular bisector.
As, , the uncertainty of the received signal vector is mainly contributed by the random noise ( ), which is Gaussian-distributed, around the signal symbol.
rr
nsr irrr
+= rr
nr
s2s1
decision boundary
d
noise distribution
( )isr rr−=
6 Oct. 2009 TELE3113 9
Digital Signal Detection
s2s1
decision boundary
d
noise distribution
Assume s1 is sent,
at the receiver, the probability of detection error is:
( )
∫
∫∞
−
∞
=
=
−>−=
2/
/
21
121
12
21
)(
)(
d
n
d/
dne
dns|nf
ssrsrP
ssP
η
πη
sent is detected is
rr
rrrrr
rr/η) n (
/η) sr (
e)s|nf()s)|s-rf(() s|rf(
e)s|rf(
2
21
1
1With
1111
1
r
rr
rrrrrrr
rr
−
−−
===
=
πη
πη
∫∞
−=x
y dyexQ 2/2
21)(πUsing and let η
ny 2=
=
= ∫∞
−
η
πη
2
21)sent is detected is (
2/
2/12
2
dQ
dyessPd
yrr
6 Oct. 2009 TELE3113 10
Digital Signal DetectionFor a signal symbol set: Misi ,...1for }{ =
r
Detection error probability is
][2
][][
11
1
i
M
i
M
kki
ik
M
iiie
sPss
Q
sPsPP
rrr
rr
sent | detection erroneous
∑∑
∑
==≠
=
−≤
=
η
If all signal symbols are equally probable, i.e. MsP i /1][ =r
sent | detection erroneous
∑∑
∑
==≠
=
−≤
=
M
i
M
kki
ik
M
iiie
ssQ
M
sPsPP
11
1
21
][][
η
rr
rr
6 Oct. 2009 TELE3113 11
Digital Signal DetectionCalculation of error probabilities:
(a) Antipodal signaling: ( ) ( )0, ;0, 21 EsEs −=+=
E
s
+
1
E
s
−
2
[ ]
=
+
=
+
≤
+=
η
η
ηη
EQ
sPsPEQ
sPEQsPEQ
sPssPsPssPPe
2
)()(2
)(2
2)(2
2
)()|()()|(
21
21
221112
detectedisdetected is
Example: for NRZ signaling which takes amplitude either +A or 0. For bit interval Tb, the energy per bit Eb=A2Tb.
signal symbol energy=E
=
=
ηηb
eEQTAQP
2
6 Oct. 2009 TELE3113 12
2 sE+
[ ]
=+
=
+
≤
+=
ηη
ηη
EQsPsPEQ
sPEQsPEQ
sPssPsPssPPe
)()(
)(22)(
22
)()|()()|(
21
21
221112
detectedisdetected is
Digital Signal Detection(b) Orthogonal signaling: ( ) ( )EsEs +=+= ,0 ;0, 21
(c) Square signaling:( ) ( ) ( ) ( )EEsEEsEEsEEs −−=+−=++=−+= , ;, ;, ;, 4321
E
s
+
1
E+
E+E−
E− 1s
3s
4s
2s
+
=
+
+
≤
=
∑
∑
=
=
ηη
ηηη
EQEQ
EQEQEQsP
sPssPP
ii
iiiie
222
2
2222
22)(
)()|detectednot is (
4
1
4
1
6 Oct. 2009 TELE3113 13
Digital Signal DetectionIntegrate-and-Dump detector
r(t)=si(t)+n(t)
≤≤−=≤≤+=
=0for 0 )(1for 0 )(
)(2
1
TtAtsTtAts
tsi
[ ]
++
=+= ∫+
0for )(1for )(
)()()(2
10
0 o
oTt
ti nta
ntadttntstzOutput of the integrator:
where
( )
∫
∫
∫
+
+
+
=
−=−=
==
Tt
to
Tt
t
Tt
t
dttnn
ATdtAa
ATAdta
0
0
0
0
0
0
)(
2
1
6 Oct. 2009 TELE3113 14
Digital Signal Detectionno is a zero-mean Gaussian random variable.
{ } { }∫∫++
==
=Tt
t
Tt
to dttnEdttnEnE
0
0
0
0
0)()(
{ } { }
{ }
( )
22
2
)()(
)(
0
0
0
0
0
0
0
0
0
0
2
22
Td
dtdt
dtdntnE
dttnEnEnVar
Tt
t
Tt
t
Tt
t
Tt
t
Tt
t
Tt
toon
o
o
o
ηεη
εεδη
εε
σ
==
−=
=
===
∫
∫ ∫
∫ ∫
∫
+
+ +
+ +
+
( ) )/()2/( 222 12
1 T
nn e
Tef on
o
o
ηασα
πησπα −− ==pdf of no:
6 Oct. 2009 TELE3113 15
Digital Signal Detection
≤≤−=≤≤+=
=0for 1for
TtAtsTtAts
tsi 0)(0)(
)(0
1
AT
s
+
1
AT
s
−
0 As
0We choose the decision threshold to be 0.
Two cases of detection error:
(a) +A is transmitted but (AT+no)<0 no<-AT
(b) -A is transmitted but (-AT+no)>0 no>+AT
Error probability:
[ ]
Tudue
APAPdT
e
dT
eAPdT
eAP
APAATnPAPAATnPP
TA
u
AT
T
AT
TAT T
ooe
ηα
π
απη
απη
απη
η
ηα
ηαηα
2 2
)()(
)()(
)()|()()|(
2
2
2
22
2
2/
)/(
)/()/(
==
−+=
−+=
−>+−<=
∫
∫
∫∫
∞ −
∞ −
∞ −−
∞−
−
Q
( )
∫
∫
=
=
=
=
∞ −
T
bb
x
u
e
dtAEEQ
duexQTAQP
0
2
2/2
2
2 2
2
Q
Q
η
πηThus,