Wireless application data has been growing rapidly. To meet the increasing demand of data rate, researchers have proposed various advanced techniques to mitigate the interference, which is a key performance bottleneck in wireless communication. However, these interference mitigation techniques in general require that the channel state information is available at the transmitter side (CSIT). In frequency division duplex (FDD) wireless systems, CSIT is obtained via pilot training from the transmitters (Tx) and CSI feedback from the receivers (Rx) and hence, acquiring CSIT results in a heavy signaling overhead. To truly enjoy the benefits of the interference mitigation techniques, it is important to take the CSIT acquisition cost into consideration and it is highly desirable to red...[ Read more ]

Wireless application data has been growing rapidly. To meet the increasing demand of data rate, researchers have proposed various advanced techniques to mitigate the interference, which is a key performance bottleneck in wireless communication. However, these interference mitigation techniques in general require that the channel state information is available at the transmitter side (CSIT). In frequency division duplex (FDD) wireless systems, CSIT is obtained via pilot training from the transmitters (Tx) and CSI feedback from the receivers (Rx) and hence, acquiring CSIT results in a heavy signaling overhead. To truly enjoy the benefits of the interference mitigation techniques, it is important to take the CSIT acquisition cost into consideration and it is highly desirable to reduce the CSIT acquisition overhead.

In this thesis, we first consider the topic of CSI feedback reduction for interference alignment (IA) in MIMO networks. To achieve this goal, three basic yet open questions need to be investigated: i) what are the possible CSI feedback reduction strategies? ii) how to quantify the CSI feedback cost? iii) IA feasibility conditions under partial CSI feedback. By answering these questions, we establish a base to achieve CSI feedback reduction for IA. We then further study into the problem of CSI feedback cost minimization subject to IA feasibility constraint. Using specific problem features, we obtain an asymptotic optimal CSI feedback solution for symmetric network topologies. From the results, we obtain simple tradeoff results between the degree of freedoms and the CSI feedback cost in MIMO networks.

Besides the above works, we further study CSIT acquisition designs in massive MIMO systems. The massive MIMO channels are usually sparse due to the limited local scattering effect between the base station and the users. Aside from this, the wireless channels further have some innate properties such as temporal correlations, or joint channel sparsity among the users due to the shared common scattering in multi-user systems. In this thesis, we shall propose compressive sensing (CS) based channel acquisition schemes, to exploit not only the channel sparsity, but also these innate channel properties (i.e., temporal correlation, joint channel sparsity), to further reduce the CSIT acquisition overhead for massive MIMO systems.