Code design for quasi-static fading channels

Today, the emergence of more demanding applications over wireless communication channels requires improvements to capacities and data rates available. The increased number of users sharing the medium of communications translates into short information frames when compared with the rate of change of the wireless channel. These types of channels are commonly referred to as quasi-static fading channels (QSFC). Although some error correction codes have been proposed for QSFCs, there has been little effort invested in trying to analyze their performance.; In this dissertation the problem of analysis of error correction codes over QSFCs is addressed. A general novel approach to performance analysis of coded systems over QSFCs is proposed. The simple case of code performance over single-input single-output (SISO) QSFC is first considered and a performance bound is derived. The derived upper bound is a much improved alternative to characterizing the performance of channel codes over QSFCs.; The case of serially concatenated space-time block codes is also analyzed. An equivalence with SISO QSFC is proved and the previously obtained results are extended to this case. When turbo-codes are used in this structure, the derived performance upper bound provided analytical grounds for the claim that unlike over AWGN channels, there is no improvement in bit error performance with increase in frame size over QSFCs.; The more challenging case of a multiple-input multiple-output QSFC is investigated next. The approach of using space-time trellis codes (STTC) to provide diversity along with error correction capabilities in a single code was analyzed. As in previous cases, the general approach proposed is used to derive performance bounds for STTCs over QSFCs. The new analysis contrasts with previous approaches that used a few error events to characterize performance of such codes. It also proved reliable in predicting how a given STTCs would fare over QSFCs. An added difficulty of STTC is that their analysis depended on an unusual eigenvalue distance spectrum which proved quite difficult to obtain so far. In this work an algorithm that proposed a soft trade-off between the effort invested in computing the spectrum and the tightness of the resulting upper bound was devised.