Digital Signal Processing Reference
In-Depth Information
N
∑
1
(
)
+
ht
+
( )
()
=
a
exp
jn
θφ
( )
t
Bt
( ),
(7.15)
k
k
k
k
n
=
where
K
NK
a
=
k
,
(7.16)
k
(
1
+
)
k
N
1
∑
( )
.
Bt
()
=
bt
()exp( )
jn
φ
t
(7.17)
k
nk
NK
(
1
+
)
k
n
=
1
It is easy to show that
B
k
(
t
) ~ C N (0,σ
2
), where σ
2
= 1/(1 +
K
k
).
Let
H
k
(
t
) denote the magnitude of
˜
k
(
t
):
N
∑
Ht ht
()
=
()
=
a
exp(
jn
(
θφ
+
( )))
t
+
Bt
()
.
(7.18)
k
k
k
k
k
n
=
1
Each user measures only
H
k
(
t
) and then determines the data rate
R
k
(
t
) at which the
data can be reliably transmitted from the base station to the user with a predetermined
SNR threshold. The determined rates
R
k
(
t
) for all the users are fed back to the base sta-
tion. The requested rate
R
k
(
t
) can be expressed as
R
k
(
t
) =
f
(
H
k
(
t
)), where
f
(·) is a non-
decreasing function, which can be assumed to be known at the base station. Hence, the
values of
H
k
(
t
) are also assumed to be known at the base station in this section.
On the right-hand side of (7.18), the first term
∑
N
(
)
+
( )
a
exp
jn
θφ
( )
t
k
k
n
=
1
is related to the LOS component and the second term
B
k
(
t
) to the diffused component.
Let
G
k
(
t
) denote the magnitude of the LOS component:
N
def
∑
θφ
1
(
)
+
( )
Gt
()
=
a
exp
jn
( ).
t
(7.19)
k
k
k
n
=
Then one can easily see that
G
k
(
t
) is maximized with ϕ(
t
) = -θ
k
and the maximum
value is
a
k
. Motivated by this observation, the following adaptive opportunistic
beamforming algorithm is proposed to improve the performance [17]:
Search WWH ::
Custom Search