• CHANNEL ESTIMATION
• DOWNLINK TRANSMISSIONS
• MC-CDMA
• NOISE POWER ESTMATION
• OFDMA
• PRE-FILTERING
• RESOURCES ALLOCATION
• TOMLINSON-HARASHIMA PRE-CODING

Future wireless communication systems are expected to support high-speed and high-quality multimedia services. In theseapplications the received signal is typically affected byfrequency-selective fading, which must be properly counteracted toavoid a severe degradation of the system performance.

MC-CDMA is a multiplexing technique that combines OFDM with direct sequence CDMA. It is robust to frequency-selective fading thanks to the underlying OFDM modulation and exploits frequency diversity by spreading the data of different users in the frequency domain. For these reasons it is considered as a promising candidate for the physical layer of future high-speed wireless communications.

Recent publications show that MC-CDMA is particularly suitable for downlink transmissions, i.e., from the base station to the mobile terminals. In these applications orthogonal spreading codes are usually employed to provide protection against co-channel interference. In the presence of multipath propagation, however, signals undergo frequency-selective fading and the code orthogonality is lost. This gives rise to multiple-access interference, which strongly limits the system performance. In the past few years several advanced multi-user detection techniques have been proposed and discussed for interference mitigation. However, in spite of their effectiveness, all these methods are quite unattractive for downlink applications since they would entail high complexity and excessive power consumption at the remote units.

As an alternative to multi-user detection, pre-filtering techniques can be employed in downlink transmissions to mitigate multiple-access interference and channel distortions. The idea behind pre-filtering is to vary the gain assigned to each subcarrier so that interference is reduced and the signal at the receiver appears undistorted. In this way, simple and low complex single-user detectors can be employed at the remote units, thereby moving most of the computational burden to the base station, where power consumption and computational resources are not critical issues.

In general terms, the main contribution of this dissertation is threefold. First, we propose and discuss several linear and non-linear pre-filtering schemes for the downlink of MC-CDMA systems equipped with multiple transmit antennas and operating in a time-division-duplex mode. The resulting schemes are derived according to different optimization criteria and aim at combating the detrimental effects of MAI while maintaining the complexity of the remote units as low as possible. A second contribution comes from providing a unified framework for investigating pre-filtering in the downlink of both MC-CDMA and OFDMA systems. The use of a unified framework comprising both MC-CDMA and OFDMA allows a fair comparison between these multiple-access technologies under the same operating conditions. It turns out that OFDMA outperforms MC-CDMA when the system resources are optimally assigned to the active users according to the actual channel realization. As we shall see, in order to work properly, all the proposed schemes require explicit knowledge of the channel responses of the active users. In time-division-duplex systems this information can be achieved by exploiting the channel reciprocity between alternative uplink and downlink transmissions. If channel variations are sufficiently slow, the channel estimates of the active users can be derived at the base station during an uplink time-slot and reused for pre-filtering in the subsequent downlink time-slot. Thus, a third contribution comes from addressing the problem of channel acquisition in the uplink of an MC-CDMA system equipped with multiple receive antennas.