. .
X X X X
X X X X
Detector
w2K
w1K
w23 w22 w21
w11 w11 w11
4 ?†‘
2 ?†‘
4 ?†‘
+
filter is operating with a sampling frequency of eight times the symbol rate. This explains the decimation
by a factor of four to produce half-symbol rate information. Once the equalization and combining
operations have completed, the STE output is decimated by a factor of two to enter symbols
to the detector chosen. This assumes the detector is a symbol rate-based detector otherwise the decimation
operation is not needed.
After collecting K consecutive samples on both antennas we can create the following column vectors.
(6.47)
Recall the MMSE weight calculation involves a covariance matrix and a correlation vector. The
covariance matrix can be obtained by stacking the received samples from both antennas into a single
column vector which results in the following MK MK square matrix.
(6.48)
which is also written in block matrix form assuming M 2 receive antennas.
(6.49)
We will show in the following sections that the covariance matrix can be written in the simplest terms below.
(6.50)
where H is the covariance matrix of the overall channel response and is the noise power. Thus the
MMSE-based weights are written as
(6.
Pages:
484
485
486
487
488
489
490
491
492
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508