EE maximization for massive MIMO with fully connected hybrid beamformers

This work deals with bit-per-Joule energy efficiency maximization in a mmWave massive MIMO link with hybrid beamforming, by transmit power and number of active RF chains optimization. The problem is formulated as a mixed-integer fractional program. By relaxation and fractional programming methods, both the optimal number of RF chains for fixed power level, and the optimal power level for fixed number of RF chains are found. Next, joint allocation is performed by alternating optimization. Numerical results indicate that in relevant scenarios, the proposed method performs like the globally optimal solution, which can be determined by exhaustive search.