A hybrid energy efficiency optimization model using machine learning for AIRS system

The energy efficiency (EE) enhancement in Unmanned Aerial Vehicles (UAVs) is critical for prolonged operations, due to the limited energy source. The Intelligent Reflecting Surface (IRS) has been integrated with UAVs in many previous works to enhance the EE. But many cases of the channel have not been sufficiently discussed in this field. This paper focuses on optimizing EE in an Aerial-IRS (AIRS) system based on machine learning (ML) and optimization algorithms in Rician channel model. By modelling the system’s placement, channel properties, and spectrum allocation, we have developed mathematical formulations for EE in AIRS and UAV-only systems. ML contributes to enhancing the optimization process, determining the best altitude to maximize EE. According to the simulation results, the AIRS system performs significantly better in terms of EE than the UAV-only system due to its improved signal-to-noise ratio and less power consumption. Furthermore, our hybrid EE optimization (HEEO) algorithm enhances gain by controlling the dynamic AIRS positions, in addition to reflected element control illustrating improvements in EE. Additionally, the hybrid optimization algorithm provides adaptability to dynamic UE positions. The EE is improved by up to 99.6% in proposed AIRS system compared to UAV only system. In addition to improving up to 4.2% when adding optimization algorithm and ML model to proposed AIRS system compared to not adding them when the horizontal distance between BS and UE is equal; 100 m. These results highlight the significance of integrating IRS into the AIRS system, and optimization algorithm with ML to improve EE in the AIRS system.