BDMA in Multicell Massive MIMO Communications: Power Allocation Algorithms

We investigate power allocation strategies for beam division multiple access transmission in multicell massive multiple-input-multiple-output (MIMO) communications. Focusing on massive MIMO downlink with only statistical channel state information at serving base stations and multiantenna terminals, the eigenmatrices of channel transmit covariance matrices become identical and independent of terminals. Utilizing these eigenmatrices as precoding matrices, we consider power allocation of beam domain transmission. By treating cochannel user interferences as noises, the objective function of achievable ergodic sum-rate of the multicell massive MIMO downlink reduces to a difference of concave functions. We identify the orthogonality conditions for optimal power allocation. The results show that the transmit power allocated to the users should be nonoverlapping across beams defined by the channel transmit covariance matrices. We present two algorithms to optimize power allocation in the beam space. We also prove the convergence of the algorithms and show that the solution obtained by the algorithms satisfies the orthogonality conditions. Numerical results confirm the efficiency and the improved performance of the proposed algorithms.