Coordinated DPC-Based Precoding Design for Energy Efficiency Optimization in Downlink Multi-Cell MIMO Systems

In this paper, we aim to maximize the total energy efficiency for a multi-cell MIMO broadcast channel with dirty paper coding, where both the base stations and users employ multiple antennas. The EE metric is defined as the ratio of the total sum-rate to the total power consumption. Because the original problem is non-convex and difficult to tackle directly, we employ the fractional programming and iterative linear approximation methods to transform it into a set of sub-problems. After using Lagrange dual decomposition, each sub-problem essentially becomes a precoding design problem in MIMO broadcast channel (BC) and is still non-convex, which is then dealt with via BCmultiple access channel (MAC) duality property. Specifically, we propose a gradient descent (GD) method to compute the uplink precoder in each MAC problem which has low complexity. Thus, the dual BC problem can be solved and each BS can iteratively update its precoding matrices with small amount of information exchange among the base stations. Numerical results validate the better performance of our proposed algorithm over conventional linear precoding method.