MIT OpenCourseWare

http://ocw.mit.edu

16.36 Communication Systems Engineering

Spring 2009

For information about citing these materials or our Terms of Use, visit: http://ocw.mit.edu/terms.
 Lectures 8 - 9: Signal Detection in Noise
and the Matched Filter

Eytan Modiano

Aero-Astro Dept.

Eytan Modiano
Slide 1
 Noise in communication systems

S(t) Channel r(t)

r(t) = S(t) + n(t) n(t)

• Noise is additional “unwanted” signal that interferes with the

transmitted signal
– Generated by electronic devices

• The noise is a random process

– Each “sample” of n(t) is a random variable

• Typically, the noise process is modeled as “Additive White

Gaussian Noise” (AWGN)
– White: Flat frequency spectrum
– Gaussian: noise distribution

Eytan Modiano
Slide 2
 Random Processes

• The auto-correlation of a random process x(t) is defined as

– Rxx(t1,t2) = E[x(t1)x(t2)]

• A random process is Wide-sense-stationary (WSS) if its mean and

auto-correlation are not a function of time. That is
– mx(t) = E[x(t)] = m
– Rxx(t1,t2) = Rx(τ), where τ = t1-t2

• If x(t) is WSS then:

– Rx(τ) = Rx(-τ)
– | Rx(τ)| <= |Rx(0)| (max is achieved at τ = 0)

• The power content of a WSS process is:

$ $
1 T /2 1 T /2
Px = E[ lim x (t)dt = lim
2
Rx (0)dt =Rx (0)
t! " T #T / 2 t !" T #T / 2

Eytan Modiano

Slide 3

 Power Spectrum of a random process

• If x(t) is WSS then the power spectral density function is given by:
Sx(f) = F[Rx(τ)]
• The total power in the process is also given by:

" "%" (
# S ( f )df = # ''& # R (t)e
! j2 $ ft
Px = x x dt * df
*)
!" !" !"

"%" (
# ''& # R (t)e
! j2 $ ft
= x df * dt
!" !"
*)

" %" ( "

# # #
= Rx (t) ' e ! j2$ft df * dt = Rx (t)+ (t)dt = Rx (0)
'&!" *)
!" !"

Eytan Modiano

Slide 4

 White noise

• The noise spectrum is flat over all relevant frequencies

– White light contains all frequencies

Sn(f)
No/2

• Notice that the total power over the entire frequency range is infinite
– But in practice we only care about the noise content within the signal
bandwidth, as the rest can be filtered out

• After filtering the only remaining noise power is that contained within the
filter bandwidth (B)
SBP(f)

No/2 No/2
-fc fc
Eytan Modiano
Slide 5
B B
 AWGN

• The effective noise content of bandpass noise is BNo

– Experimental measurements show that the pdf of the noise samples can be
modeled as zero mean gaussian random variable

1 # x 2 / 2" 2
fx (x) = e
2! "
– AKA Normal r.v., N(0,σ2)

– σ2 = Px = BNo

• The CDF of a Gaussian R.V.,

! !
% f (x)dx = %
1 #x 2 / 2' 2
Fx (! ) = P[X " ! ] = e dx
2& '
x
#$ #$

• This integral requires numerical evaluation

– Available in tables
Eytan Modiano

Slide 6

 AWGN, continued

• X(t) ~ N(0,σ2)

• X(t1), X(t2) are independent unless t1 = t2

•
#E[X(t + ! )]E[X(t )] ! " 0
Rx (! ) = E[X(t + ! )X(t)] = $
% E[X 2 (t)] ! =0

#0 !"0
=$ 2
%& ! = 0

• Rx(0) = σ2 = Px = BNo

Eytan Modiano

Slide 7

 Detection of signals in AWGN

Observe: r(t) = S(t) + n(t), t ∈ [0,T]

Decide which of S1, …, Sm was sent

• Receiver filter
– Designed to maximize signal-to-noise power ratio (SNR)

filter “sample at t=T”

y(t)
r(t) h(t) decide

• Goal: find h(t) that maximized SNR

Eytan Modiano
Slide 8
 Receiver filter

#
y(t) = r(t) * h(t) = r(! )h(t " ! )d!
0
T

#
Sampling at t = T $ y(T )= r(! )h(T " ! )d!
0
r (!) = s(! ) + n(! ) $
T T

#
y(T) = s(! )h(T " !) d! + # n(! )h(T " ! )d! = Y (T) + Y (T )
s n
0 0
2 2
%T ( %T (

Ys2 (T)
#
' s(! )h(T " ! )d! *
'& 0 *)
#
' h(! )s(T " !) d! *
'& 0 *)
SNR= = =
E[Yn2 (T )] T T

# #
N0 N0
h2 (T " t)dt h2 (T " t )dt
2 2
0 0

Eytan Modiano

Slide 9

 Matched filter: maximizes SNR

Caushy - Schwartz Inequality :

2
$ " ' " "
# * # #
2 2
g
&% !"1 (t)g2 (t)dt )( (g 1 (t) ) (g 2 (t))
!" !"

Above holds with equality iff: g1 (t) = cg2 (t) for arbitrary constant c

2
$T ' T T

#
& s(+ )h(T ! + )d+ )
&% 0 )( # #
(s(+ )) 2 d+ h 2 (T ! + )d+ T

#
2 2Es
SNR= T * 0
T
0
= (s(+ )) d+ =
2
N0 N0
# #
N0 N0
h2 (T ! t )dt h 2 (T ! t)dt 0
2 2
0 0

Above maximum is obtained iff: h(T-τ) = cS(τ)

=> h(t) = cS(T-t) = S(T-t)
Eytan Modiano
Slide 10
h(t) is said to be “matched” to the signal S(t)
 Example: PAM

Sm (t) = Amg(t), t ∈ [0,T]

Am is a constant: Binary PAM Am ∈ {0,1}

Matched filter is matched to g(t)

g(t) g(T-t) “matched filter”

A A

T T

Eytan Modiano
Slide 11
 Example, continued

Ys (t) = $ S(! )h(t " ! )d! , h(t' ) = g(T " t' ) # h(t " ! ) = g(T +! " t )
0

t t
Ys (t) = $ g(!)g(T + ! " t )d! = $ g(! )g(T " t + ! )d!
0 0

T
Ys (T ) = $ g2 (! )d!
0
Ys(t)
A2T
• Sample at t=T to obtain
maximum value

t
T
Eytan Modiano

Slide 12

 Matched filter receiver

Sample at t=kT
U(t) rx(t) g(T-t) rx(kT)

2Cos(2πfct)
Sample at t=kT
U(t) ry(t) g(T-t) ry(kT)

2Sin(2πfct)

Eytan Modiano
Slide 13
 Binary PAM example, continued

g(t)
0 => S1 = g(t)
1 => S2 = -g(t)
A

S(t) “S1(t)” T

“S2(t)”

Y(t)

“Y1(t)”
“Y2(t)”
2T 3T T 2T
T 2T
T
Eytan Modiano
Slide 14
 Alternative implementation: correlator receiver

r(t) = S(t) + n(t)

Sample at t=kT

! ()
T
r(t) Y(kT)
0

S(t)

! ! !
T T T
Y(T) = r (t)S(t) = S (t) + n(t)S(t) =Ys (T ) + Yn (T )
2
0 0 0

Notice resemblance to matched filter

Eytan Modiano
Slide 15