Energy-Efficient Power Allocation and Joint User Association in Multiuser-Downlink Massive MIMO System

Singular value decomposition is highly essential to achieve a higher performance in signal processing using massive multiple-input multiple-output (MIMO) systems. This paper aims to provide a solution to control power allocation problem identified as an essential metric in a massive MIMO system that maximizes energy efficiency (EE). The network performance was evaluated by measuring circuit power consumption to maximize EE. The computational efficiency to maximize EE power allocation is very important to fifth generation networks (5G). The study aims to maximize the non-convex EE in a downlink (DL) massive MIMO system using a proposed energy-efficient low-complexity algorithm (EELCA) that guarantees optimal power allocation solution based on Newton’s methods and joint user’s association based on the Lagrange’s decomposition method. An optimal power allocation solution in closed form to decrease the complexity of the power subject to both the maximum power and minimum data rate constrained systems was derived. Then, the unconstrained EE power allocation to solve the unconstrained optimal power was used to select the optimal power allocation by computing a root of the first derivative of the EE based on differentiating the instantaneous power allocation to maximize EE was formulated. Simulation results showed that the proposed EELCA with a total transmitted power allocation provided maximum EE for a large number of antennas at the base station (BS), Generally, non-linear schemes outperformed linear schemes. Finally, the large cost of circuit power consumption increased at the BS due to the large loss of radio frequency (RF) chains at every antenna when the signals were transmitted to all users. The maximum EE = 5.9 Mbits/joule when the number of distributed users, $K=33$ , with $\left ({p_{c}, M }\right )=\left ({ 1000\mathrm {mW},200 }\right )$ . The proposed low complexity algorithm provides the better result EE based on a training channel for a number of distributed users.