Digital Signal Processing Reference
In-Depth Information
[
Γ
v
(
ω
)]
ij
=[
Γ
DN
(
ω
)]
ij
=
sin[
ω
(
j − i
)
τ
0
]
ω
(
j − i
)
τ
0
= sinc [
ω
(
j − i
)
τ
0
]
.
(2.44)
In this scenario, the gain in SNR,
G
DN
[
h
(
ω
)], is called the directivity factor
and the directivity index is simply defined as [2], [4]
D
[
h
(
ω
)] = 10 log
10
G
DN
[
h
(
ω
)]
.
(2.45)
With diffuse noise, the filter
h
(
ω
) is often found by maximizing the direc-
tivity factor. As a result, the optimal filter is given by (2.38).
•
The noise comes from a point source at the angle
θ
N
. In this case, the
pseudo-coherence matrix is
Γ
v
(
ω
)=
d
(
ω,
cos
θ
N
)
d
H
(
ω,
cos
θ
N
)
,
(2.46)
where
T
−
ωτ
0
cos
θ
N
···e
−
(
M −
1)
ωτ
0
cos
θ
N
d
(
ω,
cos
θ
N
)=
(2.47)
1
e
is the steering vector of the noise source. We observe from (2.46) that the
pseudo-coherence matrix is singular. In fact, this is the only possibility
where the gain in SNR,
G
NS
[
h
(
ω
)], is not upper bounded and can go to
infinity. We deduce that this gain is
2
h
H
(
ω
)
d
(
ω,
cos0
◦
)
G
NS
[
h
(
ω
)] =
|h
H
(
ω
)
d
(
ω,
cos
θ
N
)
|
2
1
|h
H
(
ω
)
d
(
ω,
cos
θ
N
)
|
2
.
=
(2.48)
When the noise and desired signals come from the same direction, i.e.,
when
θ
N
=0
◦
, then there is no possible gain, i.e.,
G
NS
[
h
(
ω
)] = 1
, ∀h
(
ω
).
We also deduce the gain of an
N
th-order DMA:
1
|B
N
(
θ
N
)
|
2
.
G
NS,N
(
θ
N
)=
(2.49)
Figures 2.5, 2.6, and 2.7 depict this gain, as a function of the direction
of the noise, for the different first-order, second-order, and third-order
patterns (dipole, cardioid, hypercardioid, and supercardioid).
Search WWH ::
Custom Search