26 Data-aided timing error estimator.
It should be pointed out that an oversampled signal is used above, as long as the received signal is
sampled at twice the baud rate, then an interpolator can be used to create the further oversampled signal
to be used in the ML calculation.
Since timing drifts can occur due to uncertainties in the transmitter, channel, and receiver, it is recommended
that this estimation be repeated periodically to track such drifts. These estimation updates
can occur at frame, time slot, or subtime slot boundaries. The update rate depends on the expected worse
case timing drift. This technique can be performed blindly (directly on samples) or a priori information
such as pilot bits can be used to improve the accuracy of the estimate.
Returning our attention to the above ML timing error estimator block
diagram, we essentially have the following oversampled eye diagram and
wish to choose the best timing instance (see Fig. 6.28). Below we provide
an example where the received symbol is oversampled by a factor of 8.
As discussed above, j ( j 0, 1, . . . , 7) is used in the comparison. One
wishes to have a sampled signal sufficiently oversampled in order to meet
the performance requirements.
Pages:
508
509
510
511
512
513
514
515
516
517
518
519
520
521
522
523
524
525
526
527
528
529
530
531
532