Beamformer Design for MIMO Interference Broadcast Channels With Semidefinite Programming

We consider beamformer design for multiuser multiple-input multiple-output (MIMO) interference channels where each transmitter communicates to its intended receivers while sharing the same spectro-temporal resources with some other unintended receivers. This scenario is often referred to as MIMO interference broadcast channel (IBC) where by using interference alignment (IA), it is possible to cancel out inter and intracell interferences so that the achievable degrees of freedom could be linearly scaled up with the number of users. In this paper, we propose two distinct and novel IA algorithms, one of which is IA via signal matching (SIGMA) that revolves around interference leakage minimization and further takes the desired signal subspaces into account to output higher sum rates. The second algorithm is called bipartite rank fitting, which is founded on the prior knowledge of the rank of receive beamformers and interference matrices. Therefore, this algorithm is quite a unique approach towards interference alignment in MIMO IBCs since it performs the optimization over only receive beamformers. Furthermore, we propose a robust design of SIGMA algorithm to achieve better performance in the presence of imperfect channel state information (CSI). The key tenet of our design is that we formulate the optimization problems involved in these two algorithms as plain semidefinite programs. It is shown that the proposed algorithms are able to outperform state-of-the-art IA techniques under perfect and imperfect CSI, and the suitability of each algorithm for different network configurations is further discussed.