Also if higher-order modulation is deemed as a viable solution then the equalizer complexity
also grows [22]. One can also suggest other reasons, in any case sometimes it makes sense to
move the signal processing into the frequency domain; we will discuss this shortly [23??“25].
MI(k) aD
j1`rj (k) aJ1
i0
h^
j (i) # sh(k i) aI1
p0 aM
m1
h^
j,m(p) # sh,m(k p)`2
RECEIVER DIGITAL SIGNAL PROCESSING 317
R??“1
I+n
r(k) ??“ h(k) .sh(k) ?†
rh(k)
r2(k)
rD(k)
M(k)
sh(k)
VA ?— ?— ( )*
Estimated Desired Signal
w/Delay Spread
??“
(k)
h(k) ?†
.
.
.
FIGURE 6.21 Unified MLSE-based receiver architecture.
A unified MLSE block diagram is shown in Fig. 6.21. The received signals are grouped together
to form a vector . This received signal vector is subtracted by using the hypothesis signals. The
metric is then computed using the inverse of the interference plus noise covariance matrix.
r (k)
6.3.1 MMSE Based
A general SC-FDE-based receiver is shown in Fig. 6.22 for the case of M antennas. First the received
signal is transformed to the frequency domain through the use of a discrete Fourier transform (DFT),
next the equalizer weights are applied by multiplication. The combined signal is transformed back
into the time domain using an Inverse DFT (I-DFT) operation.
Pages:
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526