||In the past few decades, we have witnessed a significant development of wireless networks and applications, e.g., cellular systems and wireless ad hoc networks. Unfortunately, unlike its wireline counterpart, wireless communications is accompanied by some intrinsic challenges, such as randomly faded channels, which limit the achievable performance of the network. As a natural remedy, multiple-input multiple-output technology has been proposed to enhance the spectral efficiency through spatial multiplexing, and to improve the robustness and reliability of wireless links by providing spatial diversity gain. Nonetheless, in many wireless applications, the transmitters may not be able to support multiple physical antennas due to the size, complexity, cost or other constraints. Recently, a promising technique, known as cooperative communications, has been becoming increasingly popular thanks to its ability to create spatial diversity without packing multiple antennas physically into small-size mobile nodes. Given cooperative communications, many critical issues have arisen. In particular, statistical Quality-of-Service (QoS) guarantee and reliability provisioning are becoming even more important in the design of future cooperative networks. In the first part of this thesis and by considering a cooperative multi-relay network, we propose a time-slot allocation scheme to increase the system throughput subject to the queue-overflow statistical QoS requirement. Initially, assuming a block fading channel, we derive an algorithm in which each relay is allocated a time slot of optimal length during the cooperation phase. The analysis indicates that when the QoS requirement is loose, only the relay with the best average channel condition should be selected for cooperation. On the other hand, when the QoS requirement becomes more stringent, more relays should participate in cooperation. The asymptotic case when either the transmit power or the number of relays goes to infinity is discussed, and we shall reveal a tradeoff between the transmit power and the number of relays, given a target effective capacity. Then, by modeling the channel correlation by a two-state Markov model, we develop two sub-optimal time-slot allocation algorithms which can substantially increase the effective capacity compared with the opportunistic and equal allocation schemes. The results will show that the channel correlation can sharply decrease the effective capacity. In the second part of this thesis, we propose a limited feedback scheme to reduce the outage probability for a cooperative network. Specifically, based on the instantaneous conditions of the source-destination and relay-destination channels, the destination will feed back a few bits to the source indicating the transmission time of the relay. Firstly, under the assumption that the channel estimation at the receiver and the channel feedback are both perfect, we derive the outage probabilities and show that, even with only one-bit feedback, a significant improvement in terms of the outage performance can be achieved compared to the no feedback case. To simplify the expression of the outage probability in the full feedback case, a lower bound is proposed, based on which we find the sub-optimal location of the relay that can result in a close-to-optimal outage performance. Then, by analyzing the outage behavior in the asymptotically high signal-to-noise ratio regions, we demonstrate that the proposed feedback scheme can achieve the full diversity order of two. Next, in the presence of channel estimation errors, we derive the upper bounds on the outage probabilities for the proposed feedback scheme. We find that, by appropriately choosing the transmit powers for the training phase and data transmission phase, the outage performance can be greatly enhanced. Finally, we examine noisy feedback channels in the one-bit feedback case to illustrate the effect of feedback errors. It is demonstrated that, even in the presence of feedback errors, the proposed feedback scheme can still outperform the no feedback case in terms of the outage probability.