(6.67)
Ideally when the cross-correlation vector is projected on to the inverted covariance matrix, the noise subspace
contribution is zero. However, practically speaking there is residual estimation error thus causing
performance degradation. This observation leads us to the Unified eigenbased, spatio-temporal receiver
presented earlier which continually calculated the eigen spectra of the received signal.
R1 aN
s
j1
1
lj s2n
# v j # v*
j aMK
jNs1
1
sn 2
# v j # v*
j
wMMSE [H1A H1B H1C H1D HIA HIB HIC HID s2n I]1 # rd
wMMSE [H1 H2 HI1 HI2 s2n
I]1 # rd
y(k) s(k)hy1(k) s(k T)hy2(k T ) i(k)hiy1(k) i(k T )hiy2(k T ) ny(k)
x(k) s(k)hx1(k) s(k T )hx2(k T ) i(k)hix1(k) i(k T )hix2(k T ) nx(k)
310 CHAPTER SIX
RECEIVER DIGITAL SIGNAL PROCESSING 311
FIGURE 6.14 STE covariance matrix rank investigation.
1 2 3 4 5 6
STE,M = 2 Antennas, fd = 190 Hz, Fading + CCI
K = 3 Taps (T/2-Spaced), MMSE Combine
1.E??“01
1.E??“02
1.E??“03
Covariance Matrix Estimated Rank
BER
STE (T/2-Spaced), MMSE, Flat + CCI
STE (T/2-Spaced), MMSE, FSF(?„ = T/2) + CCI
STE (T/2-Spaced), MMSE, FSF(?„ = T ) + CCI
0
0.02
0.04
0.06
0.08
0.1
0.12
0.
Pages:
493
494
495
496
497
498
499
500
501
502
503
504
505
506
507
508
509
510
511
512
513
514
515
516
517