Linear beamforming and CMHP for the Gaussian MIMO broadcast channel

Abstract

In the context of network information theory, broadcast channel (BC) represents a channel where there is only one transmitter and there are a number of receivers. Broadcast channel is a suitable model for the downlink in cellular networks, wireless local area networks (WLANs), and the downstream of digital subscriber lines (DSLs). For the last 35 years, the research on broadcast channel has achieved signicant results, especially recent results on sum rate capacity and capacity region of the Gaussian multiple input multiple output broadcast channel (MIMO Gaussian BC). Those achievements have been greatly contributing to the advances in wired and wireless communications, especially in cellular networks. In this thesis, we focus on linear beamforming for the Gaussian MIMO broadcast channels in which multiple antennas at both transmitter and receivers enable the transmission of several streams to each receiver. In particular, we generalize the uplink-downlink signal-to-interference-plus-noise (SINR) duality in broadcast channel to a MIMO scenario with general cross-talk matrix. The duality is utilized to nd an optimal solution to the SINR-balancing problem in MIMO broadcast channel. An iterative algorithm is proposed to nd optimal beamforming vectors as well as optimal power allocation for the streams while maximizing the minimum of individual SINR\\`s. The results provide a general tool for characterizing the rate regions achieved by dierent coding strategies. In the second part of this thesis, we consider superposition coding and successive decoding for MIMO BC with the use of a common information. The scheme is named CMHP by some authors. We propose a lower bound to the rate region of CMHP in a 2-user MIMO BC in which each user has 2 antennas and the transmitter has 4 antennas. Numerical results are provided for illustrating the analysis, and open problems are also discussed.