| In wireless communication systems, knowledge of the channel state information at the transmitter (CSI-Tx) can often be used to improve system performance. Ideal CSI-Tx is the exact channel estimates itself, however less complete CSI-Tx such as channel statistics (channel mean and covariance) can also be useful. In closed loop methods for obtaining exact CSI-Tx, the overhead associated with the feedback from the receiver to the transmitter can be excessive for fast mobiles. However, since the statistics of the channel change much more slowly, channel-statistics-based CSI-Tx requires a much smaller overhead and is therefore attractive for use with fast mobiles. In this thesis we study ways to exploit statistics based CSI-Tx in OFDM/OFDMA systems.; Some examples of statistical CSI-Tx are antenna correlation and tap correlation. Tap correlation can arise when different paths exhibit correlated fading, or it can be caused by pulse shaping filters. We develop simple physical models for antenna and tap correlation. We show that tap correlation leads to frequency selective variations in average channel gains. Likewise, correlated angle spread arrivals lead to space selective variations in average channel gains.; Our first result is an optimal precoding approach to maximize capacity in a MIMO-OFDM system when spatial and tap correlation CSI-Tx is available. We also develop a statistical waterfilling scheme that exploits antenna and tap correlation. We show that channel correlation decreases channel capacity when no CSI-Tx is available and this is mitigated when correlation CSI-Tx is available.; The second result is an adaptive modulation (AM) approach in the presence of tap correlation for OFDM transmission. Without tap correlation CSI-Tx, uniform bit loading must be applied to all the OFDM subcarriers. We show that tap correlation hurts throughput. However, with tap correlation CSI-Tx, we can apply statistical AM. We show that statistical AM, improves throughput over uniform bit loading.; The final result in this thesis exploits tap correlation statistics in opportunistic scheduling (OS) for an OFDMA uplink. With tap correlation CSI-Tx, we propose two statistical OS schemes that schedule users to subbands with the best average channel gain. We characterize the performance of the proposed statistical OS schemes. |
