Lecture Notes Radar Systems (Week 5)

Dr Sajid Ahmed

Information Technology University

Lahore, Pakistan
e-mail: sajid.ahmed@itu.edu.pk
April 7, 2020
 Matched Filter Design (1/4)

• Consider a transmitted pulse from a radar of duration Tp .

x(t) y(t)
+ h(t)
vo(t)
v(t)
• The received signal from the target, x(t), is passed through a filter, h(t), to
maximize the output signal-to-noise ratio (SNR). The input SNR is defined as
Ps
SNRi = ,
Pv
where Ps and Pn are respectively the power of the input signal and noise.
• The output SNR can be written as
|y(to)|2
SNRo = 2 (t)}
,
{v
E o
1
 Matched Filter Design (2/4)
• The output signal at time t can be written as
Z
y(t) = h(t) ∗ x(t) = h(τ )x(t − τ )dτ (1)

• Using time and frequency domain relationship for convolution

Z
y(t) = H(f )X(f )ej2πf tdf

• The output signal at time to can be written as

Z
y(to ) = H(f )X(f )ej2πf to df

• If the noise spectral density is No /2 for all frequencies

r
No
Vo (f ) = H(f ) (2)
2
• Using Parseval’s theorem, we can write
No
Z
2
{v
E o (t)} = |H(f )|2 df
2
2
 Matched Filter Design (3/4)

• The output SNR can be written as

| H(f )X(f )ej2πf to df |2
R
SNRo = R No
2
, (3)
2
|H(f )| df
• Using Cauchy Schwarz inequality
Z 2 Z Z
2
H(f )X(f )ej2πf to df ≤ |H(f )|2 df X(f )ej2πf to df,

• The left hand term can have maximum value equal to the right hand term and
left hand terms will be equal to the right hand term iff

H(f ) = X ∗ (f )e−j2πf to (4)

• Such filter is called a Matched Filter and in time domain it can be written as

h(t) = x∗ (−t + to ) (5)

3
 Matched Filter Design (4/4)

• Using (4) in (3), the maximum value of output SNR can be found as
| X (f )e−j2πf to X(f )ej2πf to df |2
R ∗
max(SNRo ) = R No
2
|X ∗ (f )e−j2πf to |2 df
| X (f )X(f )df |2
R ∗
2E
= No R ∗ =
2 X (f )X(f )df No
• The maximum value of SNR depends on the energy of the signal and does not
depend on the shape of the signal or modulation.
• The output signal at time to when matched filter is used
Z Z
y(to ) = X ∗ (f )e−j2πf to X(f )ej2πf to df = X ∗ (f )X(f )df = E (6)

• In time domain we can write

Z Z
y(t) = h(τ )x(t − τ )dτ = x∗ (−τ + to )x(t − τ )dτ (7)

4
 Example

Draw the matched filter response if the transmitted symbol x(t) is a rectangular
pulse of duration T and amplitude 1 as shown below, if you want to maximize the
SNR at t = T .

1 1

• We know the matched filter response at time t can be written as

Z
y(t) = x∗ (−τ + T )x(t − τ )dτ

5
• y(t) at t = 0 will be

6
• y(t) at t = T /2 will be

7
• y(t) at t = T will be

8
• y(t) at t = 3T /2 will be

9
• y(t) at t = 2T will be

10
 Matched Filter of Moving Targets (1/4)

• If target is moving, there will be a frequency shift in the incoming signal.

Therefore, the filter matched to the transmitted signal will degrade the output
SNR.

• If the transmitted signal is x(t), the reflected signal from the moving target will
be x̄(t) = x(t)ej2πfD t . The Matched filter for the incoming signal, to maximise
the output SNR at to can be designed using (5) as

h(t) = x̄∗ (−t + to ),

= x∗ (−t + to )e−j2πfD (−t+to ) .

11
 Matched Filter of Moving Targets (2/4)

• The matched filter output signal at time t will be

Z
y(t) = h(τ )x̄(t − τ )dτ
Z
= x∗ (−τ + to )e−j2πfD (−τ +to ) x(t − τ )ej2πfD (t−τ ) dτ
Z
= e−j2πfD to ej2πfD t x∗ (−τ + to)x(t − τ )dτ

• If the transmitted signal is a rectangular pulse of duration T

y(t) = e−j2πfD to ej2πfD t t, for 0 ≤ t ≤ T,

= e−j2πfD to ej2πfD t (2T − t), for T ≤ t ≤ 2T.

12
 Matched Filter of Moving Targets (3/4)

• There can be an error between the Matched filter and incoming signal
frequencies.
• Assume, the reflected signal from the moving target is x̂(t) = x(t)ej2πfD t and
the matched filter signal is x̄(t) = x(t)ej2πfdt

h(t) = x̄∗ (−t + to ) = x̄∗ (−t + to)e−j2πfD (−t+to )

• The matched filter output signal at time t will be

Z Z
y(t) = h(τ )x̂(t − τ )dτ = x∗ (−τ + to )e−j2πfD (−τ +to ) x(t − τ )ej2πfd (t−τ )) dτ
Z
= e−j2πfD to ej2πfd t x∗ (−τ + to)x(t − τ )e−j2π(fd −fD )τ dτ.
Z t
= e−j2πfD to ej2πfd t e−j2π(fd −fD )τ dτ for 0 ≤ t ≤ 2T
0

13
 Matched Filter of Moving Targets (4/4)

• Solving above we can get

−j2π(fd −fD )τ t
e
y(t) = e−j2πfD to ej2πfd t
−j2π(fd − fD ) 0
−j2π(fd −fD )t
−j2πfD to j2πfd t e −1
= e e
−j2π(fd − fD )
−jπ(fd −fD )t −jπ(fd −fD )t jπ(fd −fD )t


−j2πfD to j2πfd t e e −e
= e e
−j2π(fd − fD )
−j2πfD to j2πfd t −jπ(fd −fD )t sin(π(fd − fD )t)
= e e e t
π(fd − fD )t
• At t = to the output SNR will be maximum and we get
−j2πfD to j2πfd to −jπ(fd −fD )to sin(π(fd − fD )to )
y(t = to ) = e e e to
π(fd − fD )to
• It will be a sinc(·) function in to and △f = (fd − fD ). It will be zero whenever
π(fd − fD )t = nπ or △f = t1o or to = △f 1
.
14
 Ambiguity function (1/3)

• Consider a matched filter for a waveform x(t), if the input signal from the
target is x̄(t) = x(t)ej2πfdt the output of the matched filter to maximize the
output SNR at to = 0 will be
Z
y(t, fd) = x̄(t) ∗ h(t) = X(f − fd )X ∗ (f )ej2πf tdf
Z
= x(τ )ejπfd t x∗ (−t + τ )dτ
Z
= x(τ )x∗ (−t + τ )ejπfd t dτ ≡ Â(t, fd)

• The Â(t, fd) is called complex ambiguity function and its magnitue is called a
ambiguity function.

15
 Ambiguity function (2/3)

1. At t = 0, and fd = 0
Z
Â(t = 0, fd = 0) = X(f )X ∗(f )df
Z Z
= |X(f )|2 df = E ≥ X ∗ (f − fd )X ∗(f )ej2πf t df
Z
= x(τ )x∗ (τ )dτ = γx (0) = E (8)

This property says that AF cannot be higher than at the origin.

2. The volume under the ambiguity function can be defined as
Z Z
v = |Â(t, fd)|2 dtdfd = E 2

This property states that the total volume under the ambiguity surface
(squared) equals E 2 , independent of the signal waveform.
16
3. Symmetric property
|Â(t, fd)| = |Â(−t, −fd)|

17
 Ambiguity function (3/3)

√
• Consider a pulse of amplitude 1/ τ defined between −τ /2 to τ /2. The
complex ambiguity function for it can be found as
Z τ /2−t
1 j2πfd s
Z
j2πfd s
Â(t, fd) = x(s)e x(−t + s)ds = e ds
−τ /2 τ
τ /2−t
1 ej2πfd s 1 ej2πfd (τ /2−t) − ej2πfd (−τ /2)
= =
τ j2πfd −τ /2 τ j2πfd
−j2πfd t/2 j2πfd (τ /2−t/2) j2πfd (−(τ /2−τ /2)


1e e −e
=
τ j2πfd
e−j2πfd t/2 sin(2πfd (τ /2 − t/2))
=
πfd τ

18
• The ambiguity function can be written as
sin(πfd (τ − t))
|Â(t, fd)| =
πfd τ
sin(πfd (τ ))
= at t = 0
πfd τ
|t|
= 1− at fd = 0
τ

19
 Ambiguity Function Example(1/6)

• Consider a transmitted pulse of duration τ = 10µsec.

20
 Ambiguity Function Example (2/6)

• A cut of the graph at fd = 0.

21
 Ambiguity Function Example (3/6)

• A cut of the graph at fd = 0.

0.9

0.8

0.7

0.6
|A(t,fd)|

0.5

0.4

0.3

0.2

0.1

0
-1 -0.8 -0.6 -0.4 -0.2 0 0.2 0.4 0.6 0.8 1
time 10-5
22
 Ambiguity Function Example (4/6)

• A cut of the graph at t = 0.

23
 Ambiguity Function Example (5/6)

• A cut of the graph at t = 0.

0.9

0.8

0.7

0.6
|A(t,fd)|

0.5

0.4

0.3

0.2

0.1

0
-5 -4 -3 -2 -1 0 1 2 3 4
fd 105
24
 Ambiguity Function Example (6/6)

105
0.9

4
0.8

3
0.7
2
0.6
1
fd

0 0.5

-1
0.4

-2
0.3
-3
0.2
-4

-5 0.1
-1 -0.5 0 0.5 1
Time 10-5
25