Supporting the expected high data rates required by wireless Internet and high-speed multimedia services is one of the basic requirements in broadband mobile wireless systems. However, the achievable capacity and data rate of wireless communication systems are limited by the time-varying nature of the channel. Efficient techniques for combating the time-varying effects of wireless channels can be achieved by utilizing different forms of diversity. In recent years, transmit diversity based on space-time coding (STC) has received more attention as an effective technique for combating fading. On the other hand, most existing space-time diversity techniques have been developed for flat-fading channels. Given the fact that wireless channels are generally frequency-selective, in this thesis, we aim to investigate the performance of space-time diversity schemes for wideband code-division multiple-access (WCDMA) systems over frequency-selective fading channels. The proposed receiver in this case is a rake-type receiver, which exploits the path diversity inherent to multipath propagation. Then, a decorrelator detector is used to mitigate the multiple access interference (MAI) and the known near-far problem. We derive the bit error rate (BER) expression over frequency-selective fading channels considering both the fast and slow fading cases. Finally, we show that our proposed receiver achieves the full system diversity through simulation and analytical results. Most of the work conducted in this area considers perfect knowledge of the channel at the receiver. Hence, channel identification brings significant challenges to multiple-input multiple-output (MIMO) CDMA systems. In light of this, we propose a channel estimation and data detection scheme based on the superimposed training-based approach. The proposed scheme enhances the performance by eliminating the MAI from both the channel and data estimates by employing two decorrelators; channel and data decorrelators. The performance of the proposed estimation technique is investigated over frequency-selective slow fading channels where we derived a closed-form expression for the BER as a function of the number of users, K, the number resolvable paths, L, and the number of receive antennas, V . Finally, our proposed scheme is shown to be more robust to channel estimation errors. Furthermore, both the analytical and simulation results indicate that the full system diversity is achieved. Considering that training estimation techniques suffer either from low spectral efficiency (i.e., conventional training approach) or from high pilot power consumption (i.e., superimposed training-based approach), in the last part of the thesis, we present an iterative joint detection and estimation (JDE) using the expectation-maximization (EM) algorithm for MIMO CDMA systems over frequency-selective fading channels. We also derive a closed-form expression for the optimized weight coefficients of the EM algorithm, which was shown to provide significant performance enhancement relative to the conventional equal-weight EM-based signal decomposition. Finally, our simulation results illustrate that the proposed receiver achieves near-optimum performance with modest complexity using very few training symbols.