Digital Signal Processing Reference
In-Depth Information
estimate ∆
is available by exploiting diversity in space, time, and frequency, we imple-
ment carrier frequency offset recovery (CFOR) in an adaptive closed-loop structure,
where we feed back the estimate of the frequency offset to the input of the receiver. The
CFOR reduces the time variations in the spatiotemporal propagation channel due to Δ
f
to much weaker fluctuations due to the residual δ
f
= Δ
f
- ∆
. It results in much weaker
identification errors and enables further reduction of the carrier frequency estimation
error δ
f
.
At this stage, the receiver still consists of independent modules on each subcarrier. The
purpose of the last step is to improve the performance of the overall receiver by inter-
connecting these modules and performing joint multicarrier processing. We exploit the
intercarrier correlation, intrinsic to a multicarrier system, as a type of
frequency gain
to
improve the performance by joint multicarrier channel identification and synchroniza-
tion operations. Indeed, in the context of a multicarrier system, the adjacent subcarriers
are exposed to correlated fading, especially if the delay spread of the channel is relatively
low, resulting in relatively large coherence bandwidth. Hence, averaging the adjacent
subcarrier channel parameters should improve the BER performance when transmit-
ting over such low-dispersive fading channels. Along this perspective, the parameters
common to all subcarriers can be estimated more accurately by averaging their esti-
mates over all subcarriers. These parameters include the number of multipaths, their
corresponding time delays, and the frequency offset. Other channel parameters, such
as the channel fading coefficients, are correlated but not identical over all subcarriers.
Therefore, combining them may not achieve the expected performance enhancement.
We thus introduce a moving average technique over subcarriers with high correla-
tion. The fact that subcarriers are highly correlated implies similar or identical channel
parameters over subcarriers. Yet, the noise is uncorrelated across subcarriers, and hence
the similar/common parameters can be estimated more accurately by averaging their
estimates over all subcarriers, yielding the so-called frequency gain. The variance of the
resulting estimation error is lower than the variance of the estimation error without fre-
quency gain. Please bear in mind that we used the term
frequency gain
and not
frequency
diversity
, which relies on the fact that the fading is different over different subcarriers.
10.3.2 Multicarrier Interference Subspace Rejection (MC-ISR)
Provided that an instantaneous estimate of the total interference
Î
k
,
n
=
Î
d
MAI,
k
,
n
+
Î
ICI,
k
,
n
+
achieve distortionless response to the desired signal by imposing the following simple
constraints to the combiner
W
k,n
:
H
d
d
ˆ
H
d
d
ˆ
W
U
=
1
,
WU
=
1
,
kn
,
nkn
,
,
kn
,
nkn
,
,
⇒
(10. 25)
=
H
d
d
d
ˆ
ˆ
ˆ
d
H
d
d
ˆ
WI
+
I
+
I
0
.
WI
=
0
,
kn
,
MAI,
kn
,
ICI,
k
,
n
ISI,
kn
,
kn
,
kn
,
The first constraint guarantees a distortionless response to the desired signal, while
the second directs a null to the total interference realization and thereby cancels it.
Search WWH ::
Custom Search