Adaptive Multicarrier CDMA Space-Time Receivers - Adaptive Signal Processing in Wireless Communications - page 291

spread channel vector without CFO V ˆ d ∏ ,k,n . To the best of our knowledge, we are the first
to report on and address this issue and to propose an efficient scheme for closed-loop
CFOR in a multiuser detection context. It is important to mention here that if Δ f u =
Δ f ∀ u ∈{1, …, C } (i.e., downlink), then there is no need to estimate the CFO for the
MC-ISR to null the in-cell interference. Indeed, the MC-ISR combiner W k,n satisfies the
optimization property in equation (10.25). Thus, it is not affected by the CFO of other
users, i.e.,


 


  =0.

H H

d d

d

ˆ

d

ˆ

j 2π∆ fnT

WI =⇒ 0 e WI (10. 52)

MC

kn , kn , kn , kn ,

Once the MC-ISR projection is performed in equation (10.47) after reconstruction of
Î k,n without CFO, we implement the same CFOR scheme implemented in part by equa-
tions (10.48) and (10.49). Hence, like the near-far resistant detector proposed in [41], the
multiuser CFOR problem can be transformed on the downlink into a single-user CFOR
problem, and conventional single-user methods can therefore be used to estimate the
frequency offset.*
To validate the efficiency of the proposed CFOR strategy in a multicarrier and mul-
tiuser detection scheme on the uplink, we consider a multiuser DBPSK MT-CDMA sys-
tem with seven subcarriers, a spreading factor of 96, and five in-cell users ( N c = 7, L = 96,
C = 5). We select the setup that will be introduced in section 10.4.1. The frequency offset
normalized by the subcarrier separation (Δ f × T MC ) is set to 0.005 (i.e., Δ f = 200 Hz). 15
Figure 10.4 shows the link-level results of MC-ISR with and without CFOR. Results sug-
gest that a CFO of 200 Hz has a serious impact on the performance of MC-CDMA, and
that the link-level gain with the proposed CFOR is in the range of 1 dB at a BER of 5%
before channel decoding. By comparing the link-level curves of MC-ISR with CFOR
and MC-ISR without a frequency offset (i.e., CFO=0 Hz), we notice that CFOR compen-
sates almost completely the performance loss due to the frequency offset. These results
confirm the need for and the efficiency of the proposed CFOR in a multicarrier and
multiuser detection context.

10.4 Simulation Results

10.4.1 Simulation Setup

We consider an MC-CDMA system operating at a carrier of 1.9 GHz with maximum
bandwidth of 5 MHz. We select a frequency offset Δ f of 200 Hz, the maximum error
tolerated by 3G standards† (≡ 0.1 ppm) for the frequency mismatch between the mobile
and the base station [49]. We assume a frequency-selective Rayleigh fading channel
with P u = P propagation paths with exponentially decreasing powers. The channel is

* The study of the CFOR performance is provided in [43].
† We select Δ f = 200 Hz to show that even CFO residuals below the maximum value tolerated by
3G standards result in significant losses in performance.