Pareto Optimal Energy Efficiency Power Allocation: Coping With Highly Constraints in NOMA Uplink Dense Access VLC Network

Abstract

Aiming at the heavy requirement of bandwidth and delay for immersive communication, we propose the optical heterogeneous network and focus on enhance network capacity and dense access capacity as well as energy efficiency (EE). An optical MIMO network based on NOMA is developed, and we design the power allocation optimization problem (OP) with the saturation throughput as the optimization objective. However, since the severity of the successive interference cancellation (SIC) inequality constraint are positively correlated with the number of users, the solution of OP exhibits inefficient convergence of the pareto front, and the solution of the OP often triggers a chain violation effect in the power allocation. We propose a power allocation scheme based on the Fuzzy Constraint Window Gradient Descent Differential Evolution (FCWGD-DE) algorithm. We design a FCWGD strategy to expand the population search area, together with the information sharing reproduction, the suitable factor sequence, and the proportional selection, to protect the distribution and diversity of the population, and drive the population to converge to the pareto optimal frontier while avoiding local convergence. The simulation results demonstrate that the proposed scheme achieves high EE, significantly improves system throughput, and provides stable quality of service guarantee for users.