4)
(6.5)
(6.6)
where the overall channel matrix, H is of size K (K N) and the desired signal vector, is of size
(K N) 1.
The well-known significant shortcoming of this equalizer is the noise amplifying property in the
channels that contain spectral nulls [1]. This problem is exacerbated when using the zero forcing (ZF)
criteria and eliminated for the Minimum Mean Squared Error (MMSE). The advantage of this linear
equalization technique is that it is very simple to implement and control.
6.1.2 Decision Feedback Equalizer (DFE)
The equalizer structure shown in Fig. 6.3 consists of two filters: a Feed Forward Filter (FFF) and a
Feed Back Filter (FBF). The FFF performs similar to the linear equalizer discussed above, it has as
its input the received samples. The FBF has as its input the past detected symbols that are feedback.
The purpose of the FBF is to remove that part of the ISI from the present estimate caused by the previously
detected symbols. Typically, the FFF is in fractionally spaced form and the FBF is T spaced;
this combination helps achieve symbol timing recovery insensitivity.
The DFE structure is sometimes referenced as DFE (p, q), where p is the number of FFF taps and q
is the number of FBF taps.
Pages:
466
467
468
469
470
471
472
473
474
475
476
477
478
479
480
481
482
483
484
485
486
487
488
489
490