Wireless Interference Channels with Channel Uncertainties

This thesis studies wireless interference networks with two pairs of transmitters and receivers. Each transmitter has its dedicated messages for its corresponding receiver, which may also receive undesired signal from the other transmitter. Two major challenges in wireless networks, namely, interference and channel uncertainty are addressed at the same time within the context of this simple but important type of wireless network.;This treatise can be divided into three parts. The first part studies practical receiver design for a user subject to interference from another source. The instantaneous value of the fading channel status is not available to any transmitter and receiver. The receiver can measure the channel from its desired transmitter using known pilot signals, whereas it cannot measure the channel of the interferer due to the lack of coordination and thereby lack of pilot information. In such a challenging situation, a message-passing architecture is proposed to carry out joint channel estimation, interference mitigation, and decoding. Such a design achieves a significant performance gain over conventional receivers, i.e., those with a linear or Turbo channel estimator.;The second part of this thesis studies the capacity region of fading Z-interference channels, which is a special type of interference channel where only one of the two receivers is subject to interference from the undesired transmitter. It is also assumed that the fading processes are independent over time and each realization is known at the receivers but not at the transmitters. The capacity region of such fading Gaussian Z-interference channels is characterized to within a constant gap regardless of the signal-to-noise ratios and fading statistics. In the course of this study, the capacity region of a discrete Z-interference channel with fading modeled as layered erasure is completely determined to obtain insight to the achievable scheme and converse.;In the third part of this thesis, the insights obtained from the study of Z-interference channels is applied to the analysis of MIMO interference channels, where the transmitters and receivers are equipped with multiple antennas. General independent isotropic fading is assumed, and the realization of channel fading is assumed to be known at the receivers only. The degree-of-freedom (DoF) region of such MIMO interference channels, which is a first-order approximation of the capacity region in high SNR regime, is completely determined. The results show that the entire DoF region can be achieved through independent Gaussian random coding at both transmitters. This implies that the DoF gains due to beamforming and interference alignment are completely lost in absence of channel state information at the transmitters (CSIT).;In all, this thesis contributes to the understanding of both the theoretical limits and practical use of interference networks.