1.根据UPA阵列响应的定义，有
E
{
a
t
(
θ
,
ϕ
)
a
t
H
(
θ
,
ϕ
)
}
=
E
{
(
a
t
,
v
⊗
a
t
,
h
)
(
a
t
,
v
⊗
a
t
,
h
)
H
}
=
E
{
a
t
,
v
a
t
,
h
H
⊗
a
t
,
v
a
t
,
h
H
}
=
E
{
a
t
,
v
a
t
,
v
H
}
⊗
E
{
a
t
,
h
a
t
,
h
H
}
\begin{align} & \Epsilon \{{{a}_{t}}(\theta ,\phi ){{a}_{t}}^{H}(\theta ,\phi )\} \\ & =\Epsilon \{({{a}_{t,v}}\otimes {{a}_{t,h}}){{({{a}_{t,v}}\otimes {{a}_{t,h}})}^{H}}\} \\ & =\Epsilon \{{{a}_{t,v}}{{a}_{t,h}}^{H}\otimes {{a}_{t,v}}{{a}_{t,h}}^{H}\} \\ & =\Epsilon \{{{a}_{t,v}}{{a}_{t,v}}^{H}\}\otimes \Epsilon \{{{a}_{t,h}}{{a}_{t,h}}^{H}\} \\ \end{align}
E{at(θ,ϕ)atH(θ,ϕ)}=E{(at,v⊗at,h)(at,v⊗at,h)H}=E{at,vat,hH⊗at,vat,hH}=E{at,vat,vH}⊗E{at,hat,hH}
E
{
a
t
,
v
a
t
,
v
H
}
=
1
N
t
,
v
E
{
[
1
e
j
θ
⋮
e
j
(
N
−
1
)
θ
]
[
1
e
−
j
θ
⋯
e
−
j
(
N
−
1
)
θ
]
}
=
1
N
t
,
v
E
{
[
1
e
−
j
θ
⋯
e
−
j
(
N
−
1
)
θ
e
j
θ
1
⋯
e
−
j
(
N
−
2
)
θ
⋮
⋮
⋱
⋮
e
j
(
N
−
1
)
θ
e
j
(
N
−
2
)
θ
⋯
1
]
}
\begin{align} & \Epsilon \{{{a}_{t,v}}{{a}_{t,v}}^{H}\}=\frac{1}{{{N}_{t,v}}}\Epsilon \left\{ \left[ \begin{matrix} 1 \\ {{e}^{j\theta }} \\ \vdots \\ {{e}^{j(N-1)\theta }} \\ \end{matrix} \right]\left[ \begin{matrix} \begin{matrix} 1 & {{e}^{-j\theta }} & \cdots & {{e}^{-j(N-1)\theta }} \\ \end{matrix} \\ \end{matrix} \right] \right\} \\ & =\frac{1}{{{N}_{t,v}}}\Epsilon \left\{ \left[ \begin{matrix} 1 & {{e}^{-j\theta }} & \cdots & {{e}^{-j(N-1)\theta }} \\ {{e}^{j\theta }} & 1 & \cdots & {{e}^{-j(N-2)\theta }} \\ \vdots & \vdots & \ddots & \vdots \\ {{e}^{j(N-1)\theta }} & {{e}^{j(N-2)\theta }} & \cdots & 1 \\ \end{matrix} \right] \right\} \\ \end{align}
E{at,vat,vH}=Nt,v1E⎩
⎨
⎧
1ejθ⋮ej(N−1)θ
[1e−jθ⋯e−j(N−1)θ]⎭
⎬
⎫=Nt,v1E⎩
⎨
⎧
1ejθ⋮ej(N−1)θe−jθ1⋮ej(N−2)θ⋯⋯⋱⋯e−j(N−1)θe−j(N−2)θ⋮1
⎭
⎬
⎫
可以看到
E
{
e
j
(
m
−
n
)
θ
}
=
0
\Epsilon \{{{e}^{j(m-n)\theta }}\}=0
E{ej(m−n)θ}=0,因此，（6）中的非对角元素为0，有
E
{
a
t
,
v
a
t
,
v
H
}
=
1
N
t
,
v
I
N
t
,
v
×
N
t
,
v
\Epsilon \{{{a}_{t,v}}{{a}_{t,v}}^{H}\}=\frac{1}{{{N}_{t,v}}}{{I}_{{{N}_{t,v}}\times {{N}_{t,v}}}}
E{at,vat,vH}=Nt,v1INt,v×Nt,v，现在，（1）可以得出
E
{
a
t
(
θ
,
ϕ
)
a
t
H
(
θ
,
ϕ
)
}
=
1
N
L
I
N
L
×
N
L
\Epsilon \{{{a}_{t}}(\theta ,\phi ){{a}_{t}}^{H}(\theta ,\phi )\}=\frac{1}{{{N}_{L}}}{{I}_{{{N}_{L}}\times {{N}_{L}}}}
E{at(θ,ϕ)atH(θ,ϕ)}=NL1INL×NL
2.
Θ
=
d
i
a
g
(
e
j
θ
1
,
.
.
.
,
e
j
θ
N
R
)
∈
C
N
R
×
N
R
\Theta =diag({{e}^{j{{\theta }_{1}}}},...,{{e}^{j{{\theta }_{{{N}_{R}}}}}})\in {{\mathbb{C}}^{{{N}_{R}}\times {{N}_{R}}}}
Θ=diag(ejθ1,...,ejθNR)∈CNR×NR
有
a
R
H
d
i
a
g
(
e
j
θ
1
,
.
.
.
,
e
j
θ
N
R
)
a
R
a
B
H
=
θ
d
i
a
g
(
a
R
H
)
a
R
a
B
H
{{a}_{R}}^{H}diag({{e}^{j{{\theta }_{1}}}},...,{{e}^{j{{\theta }_{{{N}_{R}}}}}}){{a}_{R}}{{a}_{B}}^{H}=\theta diag({{a}_{R}}^{H}){{a}_{R}}{{a}_{B}}^{H}
aRHdiag(ejθ1,...,ejθNR)aRaBH=θdiag(aRH)aRaBH
3.
a
B
,
l
1
H
a
B
,
l
2
=
1
N
B
α
l
1
,
l
2
e
j
β
l
1
,
l
2
{{a}_{B,{{l}_{1}}}}^{H}{{a}_{B,{{l}_{2}}}}=\frac{1}{{{N}_{B}}}{{\alpha }_{{{l}_{1}},{{l}_{2}}}}{{e}^{j{{\beta }_{{{l}_{1}},{{l}_{2}}}}}}
aB,l1HaB,l2=NB1αl1,l2ejβl1,l2
其中，
α
l
1
,
l
2
=
sin
(
1
2
(
θ
l
2
−
θ
l
1
)
N
B
,
v
)
sin
(
1
2
(
ϕ
l
2
−
ϕ
l
1
)
N
B
,
h
)
sin
(
1
2
(
θ
l
2
−
θ
l
1
)
)
sin
(
1
2
(
ϕ
l
2
−
ϕ
l
1
)
)
{{\alpha }_{{{l}_{1}},{{l}_{2}}}}=\frac{\sin (\frac{1}{2}({{\theta }_{{{l}_{2}}}}-{{\theta }_{{{l}_{1}}}}){{N}_{B,v}})\sin (\frac{1}{2}({{\phi }_{{{l}_{2}}}}-{{\phi }_{{{l}_{1}}}}){{N}_{B,h}})}{\sin (\frac{1}{2}({{\theta }_{{{l}_{2}}}}-{{\theta }_{{{l}_{1}}}}))\sin (\frac{1}{2}({{\phi }_{{{l}_{2}}}}-{{\phi }_{{{l}_{1}}}}))}
αl1,l2=sin(21(θl2−θl1))sin(21(ϕl2−ϕl1))sin(21(θl2−θl1)NB,v)sin(21(ϕl2−ϕl1)NB,h)
β
l
1
,
l
2
=
1
2
(
(
θ
l
2
−
θ
l
1
)
(
N
B
,
v
−
1
)
+
(
ϕ
l
2
−
ϕ
l
1
)
(
N
B
,
h
−
1
)
)
{{\beta }_{{{l}_{1}},{{l}_{2}}}}=\frac{1}{2}(({{\theta }_{{{l}_{2}}}}-{{\theta }_{{{l}_{1}}}})({{N}_{B,v}}-1)+({{\phi }_{{{l}_{2}}}}-{{\phi }_{{{l}_{1}}}})({{N}_{B,h}}-1))
βl1,l2=21((θl2−θl1)(NB,v−1)+(ϕl2−ϕl1)(NB,h−1))
l1=l2时，有
a
B
,
l
1
H
a
B
,
l
2
=
1
{{a}_{B,{{l}_{1}}}}^{H}{{a}_{B,{{l}_{2}}}}=1
aB,l1HaB,l2=1
l1≠l2时，
a
B
,
l
1
H
a
B
,
l
2
→
0
{{a}_{B,{{l}_{1}}}}^{H}{{a}_{B,{{l}_{2}}}}\to 0
aB,l1HaB,l2→0
4603
被折叠的 条评论
为什么被折叠?
到【灌水乐园】发言
