Joint interference alignment and power allocation under perfect and imperfect CSI

We present centralized iterative algorithms that jointly determine the optimal transmit and receive filters as well as the optimal power allocation for a K-user multiple-input multiple-output (MIMO) interference channel (IC). The optimality criterion is based on the achievable sum-rate and the average per user multiplexing gain in the MIMO IC. By allowing channel state information (CSI) exchanged between base stations (BSs) and a central unit (CU), we design a feedback topology where CU collects local CSIs from all BSs, computes all transmit and receive filters and sends them to corresponding user-BS pairs. Note that the local CSIs at BSs are obtained from the estimation of the channel states during the so- called uplink-training phase. At the CU, we propose iterative algorithms utilizing alternating optimization strategy to design the filters. In most of the studies on the MIMO IC, choice of equal transmit powers for all user-BS pairs ignores the essential need to search for the optimal power allocation policy; they do not take the full advantage of the system's total power. Thus, how to allocate power among all the user-BS pairs in the network based on the sum-rate maximization strategy and under a sum power constraint is another key to this paper.